State of the art in electric batteries’ State-of-Health (SoH) estimation with machine learning: a review

Academic Editor: Jose Luis Calvo-Rolle Received: 13 January 2025 Revised: 31 January 2025 Accepted: 4 February 2025 Published: 6 February 2025 Citation: Sylvestrin, G.R.; Maciel, J.N.; Amorim, M.L.M.; Carmo, J.P.; Afonso, J.A.; Lopes, S.F.; Ando Junior, O.H. State of the Art in Electric Batteries’ State-of-Health (SoH) Estimation with Machine Learning: A Review. Energies 2025,18, 746. https://doi.org/10.3390/ en18030746 Copyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/). Review State of the Art in Electric Batteries’ State-of-Health (SoH) Estimation with Machine Learning: A Review Giovane Ronei Sylvestrin 1,2, Joylan Nunes Maciel 1,2 , Marcio Luís Munhoz Amorim 3, João Paulo Carmo 3, José A. Afonso 4,* , Sérgio F. Lopes 5and Oswaldo Hideo Ando Junior 1,2,6,* 1Interdisciplinary Postgraduate Program in Energy & Sustainability (PPGIES), Federal University of Latin American Integration—UNILA, Paraná City 85867-000, PR, Brazil; [email protected] (G.R.S.); [email protected] (J.N.M.) 2Research Group on Energy & Energy Sustainability (GPEnSE), Academic Unit of Cabo de Santo Agostinho (UACSA), Federal Rural University of Pernambuco (UFRPE), Cabo de Santo Agostinho 54518-430, PE, Brazil 3Group of Metamaterials Microwaves and Optics (GMeta), Department of Electrical Engineering (SEL), University of São Paulo (USP), Avenida Trabalhador São-Carlense, Nr. 400, Parque Industrial Arnold Schmidt, São Carlos 13566-590, SP, Brazil; [email protected] (M.L.M.A.); [email protected] (J.P.C.) 4Center for Microelectromechanical Systems (CMEMS), University of Minho, 4800-058 Guimarães, Portugal 5Centro Algoritmi/LASI, University of Minho, 4704-553 Guimarães, Portugal; ser[email protected] 6 Smart Grid Laboratory (LabREI), Center for Alternative and Renewable Research (CEAR), Federal University of Paraiba (UFPB), João Pessoa 58051-900, PB, Brazil *Correspondence: [email protected] (J.A.A.); [email protected] (O.H.A.J.) Abstract: The sustainable reuse of batteries after their first life in electric vehicles requires accurate state-of-health (SoH) estimation to ensure safe and efficient repurposing. This study applies the systematic ProKnow-C methodology to analyze the state of the art in SoH estimation using machine learning (ML). A bibliographic portfolio of 534 papers (from 2018 onward) was constructed, revealing key research trends. Public datasets are increasingly favored, appearing in 60% of the studies and reaching 76% in 2023. Among 12 identified sources covering 20 datasets from different lithium battery technologies, NASA’s Prognostics Center of Excellence contributes 51% of them. Deep learning (DL) dominates the field, comprising 57.5% of the implementations, with LSTM networks used in 22% of the cases. This study also explores hybrid models and the emerging role of transfer learning (TL) in improving SoH prediction accuracy. This study also highlights the potential applications of SoH predictions in energy informatics and smart systems, such as smart grids and Internet-of-Things (IoT) devices. By integrating accurate SoH estimates into real-time monitoring systems and wireless sensor networks, it is possible to enhance energy efficiency, optimize battery management, and promote sustainable energy practices. These applications reinforce the relevance of machine-learning-based SoH predictions in improving the resilience and sustainability of energy systems. Finally, an assessment of implemented algorithms and their performances provides a structured overview of the field, identifying opportunities for future advancements. Keywords: state of health; battery; machine learning; ProKnow-C; public datasets; energy informatics; smart grids; internet of things; deep learning 1. Introduction The worldwide increase in battery usage is evident in various fields, especially in electric vehicles. Research efforts aim to improve battery efficiency, extend lifespan, and reduce charging time, driven by the demands of a growing global market. Alongside Energies 2025,18, 746 https://doi.org/10.3390/en18030746 Energies 2025,18, 746 2 of 77 advancements in technology, vehicle battery reuse has emerged as a key area of focus. Batteries can serve automotive purposes until their capacity drops to about 80% of the nominal value. Beyond this point, replacement is necessary to meet the power requirements of vehicles [ 1 ]. However, the cells from these batteries can still be repurposed for other applications, such as stationary energy storage systems connected to photovoltaic generation devices. This process, known as second use, offers a sustainable way to extend battery life. The repurposing of a second-use battery is still a process that requires improvement because of construction and safety difficulties. Accurate estimation of batteries’ SoH is pivotal in advancing sustainable energy solutions. By integrating SoH predictions into smart grids and IoT systems, it is possible to optimize energy management, enhance system resilience, and reduce waste, aligning with broader energy informatics and sustainability goals. With increasing computational advances, the presence of smart sensors, and the era of big data, there has been growing research interest in applying machine-learning (ML) algorithms of artificial intelligence [ 2 – 4 ]. Accurate SoH characterization is essential for assessing cells suitable for reuse. As new datasets become available, the volume of research connecting ML to SoH estimation continues to grow, as demonstrated by numerous recent studies [5–9]. The estimation of the SoH of batteries is a critical step for enhancing their lifecycle management, especially in applications where reliability and performance are paramount [ 1 ]. Commonly employed methods for SoH estimation can be broadly classified into electrochemical approaches and model-based, data-driven, and hybrid methods [ 10 – 13 ]. Electrochemical approaches, although less common in operational environments because of their invasive nature, offer unparalleled precision for understanding battery degradation mechanisms. For instance, differential voltage analysis and differential capacity analysis [ 14 , 15 ] are used to track specific aging signatures by analyzing voltage–capacity profiles during charge/discharge cycles. These methods, combined with techniques like cyclic voltammetry [ 16 ] or advanced electrochemical impedance spectroscopy [ 17 ], provide detailed insights into phenomena such as lithium plating and active material loss. Although such methods are typically applied in laboratory settings, recent advances in sensor technology and signal processing aim to make them more feasible for real-time SoH estimation [15]. Model-based methods rely on electrochemical or equivalent-circuit models to predict the SoH by capturing the physical and chemical behaviors of the battery [ 18 ]. Techniques such as electrochemical impedance spectroscopy [ 19 ], Kalman filtering [ 18 ], and particle filtering [ 20 ] are widely used in this category. These methods offer precise insights into battery performance but often require complex parameter tuning and are computationally intensive [18,20]. Data-driven methods, on the other hand, utilize ML and DL algorithms to analyze large datasets and uncover patterns indicative of battery degradation [ 10 ]. These methods excel in modeling nonlinear relationships and adapting to diverse battery chemistries and usage patterns [ 10 , 11 ]. However, their reliance on large amounts of labeled data and challenges in interpretability limit their direct application in some scenarios [10,11,13]. Hybrid methods combine the strengths of model-based and data-driven approaches, leveraging physical models to enhance the interpretability and robustness of ML-based predictions. In [21,22], hybrid approaches integrating equivalent-circuit models with ML techniques are proposed, achieving a balance between accuracy and computational complexity. Despite their promise, hybrid methods often require significant domain-specific expertise and extensive computational resources [23]. Recent advancements in computational power and the availability of large datasets have significantly boosted the prominence of data-driven methods, making them a corner- Energies 2025,18, 746 3 of 77 stone in SoH estimation for large-scale applications, like electric vehicles and stationary energy storage systems [10,11]. Nonetheless, challenges persist regarding data availability, algorithmic generalization, and system interpretability [13]. The work presented in [ 10 ] provides a list of advantages and disadvantages of using ML algorithms, highlighting the need for open platforms for data sharing and modeling techniques as a necessary step for the advancement of the research field. In the context of challenges and prospects, studies [ 24 – 26 ], which focus on exploring DL techniques for estimating the remaining battery life, are also noteworthy, while in [ 27 ], this theme is reviewed from the perspective of transfer-learning usage. In [ 28 ], challenges and prospects are addressed considering the importance of feature extraction, construction, and selection for health state modeling. The importance of battery health characterization is presented in [ 12 ], under the challenges of scaling second-use batteries. In [ 11 ], a relevant review of state-of-charge (SoC) and -health estimation is presented, where the authors reveal comparative results mainly considering neural networks, such as feedforward neural networks (FFNNs), recurrent neural networks (RNNs), and long short-term memories (LSTMs). Studies [ 29 – 31 ] also provide a review focused on comparing techniques for studying battery degradation. In all the relevant review papers in recent years that have been analyzed, a common gap can be pointed out: the absence of a structured methodology that underpins the analysis portfolio and leads to the authors’ conclusions. Although significant research has been conducted on data-driven algorithms for SoH estimation, systematic methodologies are lacking to ensure the selection of highly relevant studies for constructing a reliable state-of-the-art overview. The absence of such approaches makes it difficult to identify emerging trends in the field. In this context, and given the relevance of the topic, this work aims to explore the recent state-of-the-art panorama, from the last 5 years, for the estimation of batteries’ SoHs. To achieve this, we start with the explanation and demonstration of a structured and systematic methodology, known as ProKnow-C (Knowledge Development Process Constructivist) [ 32 ], to obtain a centralized bibliographic portfolio on the topic of predicting the health states of batteries, using ML. ProKnow-C is a structured process designed to assist researchers in systematically identifying, selecting, and analyzing a bibliographic portfolio aligned with their research objectives [ 32 – 34 ]. It stands out as a comprehensive approach for conducting literature reviews because it combines quantitative and qualitative criteria, ensuring the inclusion of highly relevant and impactful studies while minimizing biases often present in manual selection processes [ 32 , 35 ]. By applying this methodology, we aim to construct a robust bibliographic portfolio that provides a reliable foundation for evaluating the state of the art in batteries’ SoH estimations. In this way, this paper’s contributions are summarized as follows: Application of the ProKnow-C Methodology: The presentation and demonstration of the ProKnow-C systematic methodology for building a bibliographic portfolio. This systematic approach allows for rigorous and structured literature reviews. Characterization of the Research Scenario (State of the Art): Characterization of the current scenario of studies in the health-state estimation of batteries, using ML by analyzing 534 relevant articles published between 2018 and 2024. This provides a comprehensive state of the art in the current research field. Public Dataset Compilation: Presentation, detailing, and summary of 20 public datasets from 12 different sources, primarily from university research centers, for selecting suitable datasets for research in SoH, SoC, and battery energy storage systems. Machine-Learning and Deep-Learning Algorithms: Research of the main ML algorithms used in studies predicting response variables related to battery degradation, including deep-learning, hybrid, and transfer-learning models. Energies 2025,18, 746 4 of 77 Performance Analysis of SoH Estimation Models: A comprehensive performance analysis of state-of-health estimation models, focusing on various response variables, including SoH, remaining-useful-life, current-lifecycle, capacity, trajectory, and earlyuseful-life predictions. The comparison involves 21 studies, allowing for both fair and broader comparisons. First Study Applying ProKnow-C to Batteries’ SoHs: This study is the first to apply the ProKnow-C systematic review method in the context of batteries’ state-of-health estimations. This pioneering application sets a new standard for structured literature reviews in this field. The remainder of this paper is organized as follows: Section 2details a systematic review using the ProKnow-C methodology, enabling rigorous and structured literature selection, mapping the state of the art in studies and experiments and available open datasets for the applicability of these techniques. Section 3discusses in detail the content analysis of the bibliographic review, as well as the analyses of the papers composing the bibliographic portfolio on this topic, including the main databases and ML algorithms, along with the key literary studies. Additionally, this section explores the potential practical applications of SoH estimation in energy informatics, smart grids, and IoT systems, highlighting its role in enhancing energy efficiency, sustainability, and operational resilience. Finally, Section 4offers concluding remarks, summarizes the main results, and provides suggestions for future research, exploring potential developments based on the integration of artificial intelligence in various scenarios while identifying gaps and opportunities for future research. 2. Systematic Review The adoption of systematic processes for bibliographic surveying allows for optimizing the quality of the material obtained on a particular topic, as it makes the process more analytical and rigorous, thereby improving the reliability of the results found. As this is an initial and fundamental stage for all research development, methods that increase the robustness of a bibliographic portfolio are essential [36]. In this paper, the systematic method ProKnow-C is employed to obtain a recent and scientifically relevant bibliographic portfolio on the use of ML in estimating SoHs of batteries. ProKnow-C was developed at the Laboratory of Multicriteria Methodologies in Decision Support (LabMCDA) at the Federal University of Santa Catarina (UFSC) and patented in 2010 [ 32 ]. This method has been applied in research in various areas, and some examples of ProKnow-C applications can be observed in [33,35,36]. Within the field of electric batteries, the ProKnow-C method was applied in [ 34 ] to define the state of the art in lithium-ion-battery recycling. Although related, the present study specifically focuses on analyzing the state of the art in SoH estimation using ML methods. To date, no similar study applying ProKnow-C has been observed. The ProKnow-C method consists of four main stages [36]: • Selection of a portfolio of papers on the research topic: This involves defining research keywords, searching in databases, and filtering articles based on alignment with the research objective, citation metrics, and relevance; • Bibliometric analysis of the portfolio: This stage examines scientific indicators, such as the number of articles, citation counts, authors, and journals, to assess the portfolio’s comprehensiveness and scientific impact; • Systemic analysis: The selected articles are deeply analyzed for insights and patterns and the identification of possible research gaps; Energies 2025,18, 746 5 of 77 • Definition of the research question and objective: The results from the previous stages are synthesized to refine the scope and formulate precise research questions and objectives. This paper presents the results of the first three stages of the ProKnow-C method, along with the analysis of the selected relevant papers. These stages represent a comprehensive state-of-the-art review of the broader research field, serving as a basis for refining the focus to a more specific and well-defined niche. Other well-known systematic review methods can be found in the literature and may be used as alternatives to ProKnow-C. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [ 37 ] emphasizes transparency and replicability through strict adherence to predefined inclusion and exclusion criteria, making it widely regarded as a gold standard in fields such as energy systems, environmental science, artificial intelligence, and other technical domains [ 38 – 40 ]. However, PRISMA does not include a bibliometric evaluation phase or tools for multicriteria decision-making, which are central to ProKnow-C. Similarly, SALSA (Search, Appraisal, Synthesis, and Analysis) [ 41 ] focuses more on synthesizing and analyzing evidence but lacks the portfolio alignment capabilities of ProKnow-C, which ensures a targeted and relevant selection of articles. Another method, Scoping Reviews, is designed to map the breadth and depth of the literature on a topic, making it well-suited for exploratory studies or identifying gaps in the literature [ 42 ]. Although Scoping Reviews provides a broad overview, it is less structured in terms of bibliometric evaluation and often does not employ multicriteria tools to refine the portfolio, which are key strengths of ProKnow-C [43]. However, as with any method, ProKnow-C has its limitations. The subjective alignment analysis stage, although useful for tailoring the portfolio to specific objectives, may reduce repeatability [ 35 ]. Additionally, its reliance on citation metrics might overlook emerging but seldom-cited studies [35]. 2.1. Bibliographic Portfolio Selection This section describes the selection of the bibliographic portfolio, initially, the set of axes and keywords that encompass the theme of this research, i.e., the use of ML in estimating SoH, was defined. As shown in Table 1, axis 1 corresponds to the study object, which is batteries. Axes 2 and 3 encompass terms related to the definition of SoH and its estimation, respectively. Axis 4 includes terms related to artificial intelligence algorithms, machine learning, deep learning, and ensembles. Table 1. Research axes for the bibliographic portfolio selection. Axis 1 Axis 2 Axis 3 Axis 4 battery state of health estimation machine learning cycle life prediction neural network lifetime features transfer learning aging second use artificial intelligence degradation boosting useful life quantile regression ensemble deep learning Energies 2025,18, 746 6 of 77 The axes were combined using the conditional logic AND, resulting in 192 combinations searched in the Scopus database. Filters were established for documents of the types of papers and reviews, searching for the keywords in titles, keywords, and abstracts, as well as defining a research horizon of publications of up to 5 years old. The Scopus database was selected because of the larger volume of papers returned compared to other databases, such as Web of Science, as well as the presence of journals focused on areas possibly related to the research. An example of a condition resulting from the combination of axes was (TITLE-ABS-KEY(battery) and TITLE-ABS-KEY(state of health) and TITLEABS-KEY(prediction) and TITLE-ABS-KEY(neural network) and PUBYEAR > 2017 and (LIMIT-TO (DOCTYPE, “ar”) OR LIMIT-TO (DOCTYPE, “re”)). Table 2presents the adherence metrics for the keywords used in the combinations of axes. The percentages shown quantify the portions of the total number of raw articles in which a particular keyword was included in the performed combinations. Within axis 2 combinations, a higher adherence rate to the term “state of health” is observed, while in axis 3 and 4 combinations, the keywords “prediction” and “neural network” stand out, respectively. These adherence metrics suggest that within the theme of research related to battery’s state of health, the term “state of health” tends to be more applied, often in connection with “prediction” studies utilizing “neural networks”. It is important to note that although some terms show low adherence rates, they remain relevant for identifying potentially important papers that may explore emerging trends in an area of research still underexplored. Table 2. Adherence to the research axes. Axis Keyword Keyword Adherence Rate Axis 2 state of health 29.3% degradation 20.6% aging 17.7% useful life 13.6% cycle life 10.9% lifetime 7.9% Axis 3 prediction 36.5% estimation 30.9% features 24.9% second use 7.8% Axis 4 neural network 36.9% machine learning 28.6% deep learning 14.6% ensemble 6.2% transfer learning 5.5% artificial intelligence 4.5% boosting 3.4% quantile regression 0.3% This search was conducted on 14 January 2024, resulting in a total of 6032 papers (with 275 papers from 2024). Although there were papers from 2024, for the calculation of a publication horizon of up to 5 years, research from 2018 onward was considered, thus having 6 complete years of publications for analysis plus two weeks of publications in 2024. The papers were exported in “.csv” format in each iteration of the 192 combinations of axes. Energies 2025,18, 746 7 of 77 The initial flow proposed by ProKnow-C is presented in Figure 1. The objective of this first stage is to significantly reduce the volume of papers in the RPD (raw paper database) obtained from the combinations of research axes. To achieve this, filters are applied to perform a preliminary selection of articles related to the research theme. The following filters are applied: Figure 1. Flow I for obtaining the bibliographic portfolio. Redundancy Filter: This is the first step of ProKnow-C, where the RPD papers are analyzed for duplication. In this stage, the “.csv” files resulting from the axis combinations were processed through a Python script that performed concatenation and removal of duplicates according to the title and publication year fields. A total of 4682 samples were removed from the RPD. Title Alignment Filter: This involves reading the papers’ titles to assess whether they are aligned with the research theme, as identified by the researchers. Out of the 1350 papers remaining after the previous filter, 722 were deemed to be not aligned with the research. Among the papers not selected were studies focused on SoH analysis in electrochemical contexts and laboratory experimental phases, which are considered as preliminary steps before exploring databases and implementing ML models. Scientific Recognition Filter: This step involves analyzing the number of citations within the RPD. In this step, the remaining 672 papers are sorted in descending order by citation count. According to the cumulative percentage of citations and a predefined cutoff percentage, the portfolio is divided into two repositories: K and P. The K repository consists of papers considered as scientifically recognized, containing 80% of the citations in the input portfolio for this filter, totaling 85 publications in the case study. The cutoff percentage is determined by the researchers, with [ 32 , 36 ] recommending a range between 70% and 90%. The P repository comprises 543 papers exceeding the defined cutoff threshold. The second flow of article selection steps for the bibliographic portfolio, using ProKnow-C, is presented in the flowchart in Figure 2. In this second phase of the method, the objective is to verify the alignment of the papers remaining from the first phase with the content presented in their abstracts. The following steps are applied: Energies 2025,18, 746 8 of 77 Figure 2. Flow II for obtaining the bibliographic portfolio. K Repository Alignment: The abstracts of the papers considered as scientifically recognized are analyzed by the researcher(s) to determine whether the research aligns with the intended research objectives. If the article is aligned, it remains in the ProKnow-C flow; otherwise, it is excluded. The 85 articles in repository K were considered as aligned with SoH estimations using ML. Creation of the Author Database (AD) and Repository A: This step involves identifying the authors of the papers approved in the previous step and creating a database of authors deemed as relevant to the research theme. The selected papers are considered as aligned with the research theme and form repository A, which constitutes the first part of the final portfolio. P Repository Alignment: This step analyzes the papers that did not reach the level of scientific recognition. These papers are divided into two categories based on their year of publication. Articles published more than two years ago are pre-selected if one of their authors is present in the AD corresponding to repository A. If no match is found in the author database, the papers are excluded. The remaining papers are then evaluated for abstract alignment, and if the expected alignment is confirmed, they are approved in the flow and included in repository B. Recent articles are not evaluated based on the AD; instead, their abstracts are directly analyzed for alignment, and approved papers are added to repository B. In this case study, in the initial analysis of repository P, which contained a total of 543 papers, 518 were recent publications from the past two years. Of the twenty-five articles older than two years, eighteen were excluded because of the absence in the AD, and the remaining seven were added to the group containing the 518 recent papers. Of the 525 papers analyzed in this stage, 449 were found to be aligned and formed repository B. Creation of Repository C: This step involves combining repositories A and B to form the final bibliographic portfolio resulting from the application of ProKnow-C. The final portfolio comprises papers aligned with the research theme, including scientifically recognized studies in terms of citations, recent articles with potential, and publications by researchers deemed as relevant to the field. The union of repositories A and B forms the final bibliographic portfolio with 534 papers, representing about 9% of the initial raw paper portfolio. From this group, a high degree of alignment with the research is expected, along with the ability to describe the current state of the art, serving as a basis for the development of the target research. Table 3presents the 40 most relevant papers in terms of the number of citations in the final portfolio. This number is based on the recommendation from [ 32 ] to evaluate an ideal vol- Energies 2025,18, 746 9 of 77 ume of between 20 and 40 papers. However, each field of research and development phase has its own characteristics that influence the ideal volume of papers. Because this work aims to reveal the current research scenario within SoH estimation using ML, a portfolio approximately ten times larger than the volume recommended by [ 21 ] was constructed to enable more robust inferences regarding the algorithms employed, datasets used, and performances achieved. Table 3. Top 40 papers from the bibliographic portfolio. Title Citations Ref. Data-driven prediction of battery cycle life before capacity degradation 1453 [1] Long short-term memory recurrent neural network for remaining-useful-life prediction of lithium-ion batteries 880 [44] Data-driven health estimation and lifetime prediction of lithium-ion batteries: A review 749 [10] A data-driven approach with uncertainty quantification for predicting future capacities and remaining useful life of lithium-ion batteries 434 [45] Predicting the states of charge and health of batteries using data-driven machine learning 405 [46] Random forest regression for online capacity estimation of lithium-ion batteries 398 [47] Remaining-useful-life prediction for lithium-ion batteries based on a hybrid model combining the long short-term memory and Elman neural networks 316 [48] Remaining-useful-life prediction for lithium-ion batteries: A deep-learning approach 313 [49] A data-driven auto-CNN-LSTM prediction model for lithium-ion-batteries’ remaining useful life 291 [50] State-of-health estimation and remaining-useful-life prediction for the lithium-ion battery based on a variant long short-term memory neural network 284 [51] Machine learning applied to electrified-vehicle-batteries’ state-of-charge and state-of-health estimations: State of the art 267 [11] Modified Gaussian process regression models for cyclic capacity prediction of lithium-ion batteries 262 [52] A deep-learning method for online capacity estimation of lithium-ion batteries 260 [53] Machine-learning pipeline for batteries’ state-of-health estimations 246 [54] A neural-network-based method for RUL prediction and SOH monitoring of lithium-ion batteries 245 [55] A novel estimation method for the state of health of lithium-ion batteries using a prior-knowledge-based neural network and a Markov chain 239 [56] Energies 2025,18, 746 16 of 77 Figure 9. Distribution of the number of publications by journal in the bibliographic portfolio. 2.2.4. Relevance of Keywords The bibliographic portfolio consists of 2753 keywords, of which 1044 are unique. Figure 10 presents the distribution of the keywords, where a significant concentration is observed for the terms “lithium-ion battery” and “state of health”, which are, indeed, the objects and central themes of this research. The prominence of “state of health” reinforces its position as a pivotal concept in this study, guiding much of the research efforts in this field. Additionally, “machine learning” appears as the fifth most frequent term, reflecting the critical role of artificial intelligence in advancing battery SoH estimation. The term “remaining useful life” is also notable, corresponding to one of the main response variables in the study of the SoH. Regarding techniques, the frequent appearances of “LSTM neural networks” and “deep learning” highlight the increasing adoption of advanced computational models. This reflects the growing sophistication in predictive analytics, as researchers seek more accurate and robust approaches to model battery degradation. Figure 11 depicts the distribution of keyword connections within the selected papers. The pattern of terms mirrors the previous distribution, showing how central keywords, such as “state of health” and “machine learning”, branch out across diverse contexts. This interconnectedness illustrates the multidisciplinary nature of SoH research, bridging fields like energy systems, artificial intelligence, and sustainability. Energies 2025,18, 746 17 of 77 Figure 10. Distribution of keywords in the bibliographic portfolio. Figure 11. Number of connections of keywords within the bibliographic portfolio. Figures 10 and 11 together emphasize the importance of keywords in structuring and advancing the field. Although the dominant terms reflect the current research focus, the variety and connections among keywords indicate the evolving boundaries of the field and its responsiveness to emerging challenges and technologies. 3. Content Analysis The bibliographic portfolio of 534 papers, following the ProKnow-C methodology analyzed in the previous section, was explored to characterize the current scenario in the field of SoH estimation in batteries. First, we outline the survey of public databases found in the bibliographic portfolio, followed by the techniques and algorithms implemented in the papers. The algorithms are further analyzed according to categories of modeling, Energies 2025,18, 746 18 of 77 including DL, algorithmic hybridization, and transfer learning. A situational overview of the performance is highlighted, and key review studies in the field are analyzed. This section concludes with implications for energy informatics and intelligent systems. 3.1. Portfolio Overview Except for review papers, the objectives of the studies are directly related to establishing various forms, either algorithmically or through different approaches, to perform SoH estimation. In the presentation of the most cited papers in Table 3, it is noted that there is a significant focus on testing different types of ML algorithms and how their methods can increase the accuracy of SoH estimation. This is the case with the study presented in [ 44 ] with long short-term memory (LSTM) neural networks, the combination of LSTM and extreme-learning-machine (ELM) neural networks presented in [ 48 ], or the use of ensemble methods, such as bagging in decision trees, through the random forest (RF) algorithm in [ 47 ]. In summary, papers with the highest number of citations in the portfolio typically focus on increasing estimation accuracy through experiments related to algorithms. Compatible studies can also be highlighted, such as the study presented in [ 78 ], which compares the LSTM, FNN, and CNN neural network algorithms, where the LSTM algorithm achieved the best performance, with an MAPE (mean absolute percentage error) of around 0.5%, compared to about 1.5% for FNN and 2% for CNN networks. In [ 79 ], the authors compare different probabilistic and time-series algorithms (ARIMAX, linear quantile regression, bootstrap multiple linear regression, and Bayesian bootstrap multiple linear regression), with the best results obtained from the Bayesian bootstrap multiple linear regression algorithm’s quantile regression, which achieved MAPE values ranging from 0.2% to 1%. Other highlighted comparative studies include [66,80]. The concept of feature engineering, which includes manipulation and selection, is a relevant theme presented in the portfolio, as much of the algorithm’s performance lies in the stress of creating features that provide discrimination for predictions. Study [ 81 ] introduces an autonomous feature selection method, which is based first on an initial selection considering correlation coefficients, tree algorithms, and a variance factor, followed by an iterative method for feature combination. In [ 82 ], an analysis of feature engineering for SoH estimation is presented, evaluating different techniques for feature creation and selection, such as univariate selection by Pearson correlation, feature importance, feature clustering, genetic algorithms, and sequential feature selection. The authors concluded that the use of sequential selection presented a good balance between performance and computational cost. Feature tests were evaluated using SVM and ExtraTree-based algorithms. Other studies focusing on features can be consulted in [81,83–96]. The quest for automating modeling processes was found in the development of an autoML approach in [ 97 ]. The framework built is capable of performing the entire modeling cycle using Bayesian optimization, eliminating the need for researchers from other fields to spend time on laborious steps, such as feature extraction, construction, and selection. The results obtained yielded MAEs (median absolute errors) ranging from about 0.02% to 0.05% for SoH estimation. Interpretability model analyses were found in studies [ 98 , 99 ], based on the use of a technique known as SHAP (Shapley Additive Explanations). SHAP is based on a theoretical game approach that seeks to explain the output of any ML model by quantifying how each feature impacts the model’s prediction [ 100 , 101 ]. Other model interpretability approaches were also addressed in [102–105]. Approaches related to hyperparameter tuning were also found in the portfolio. In [ 106 ], the authors explore the Bayesian optimization of hyperparameters in a combination of DCNN and LSTM neural networks, achieving an RMSE (root-mean-square error) of 0.0061 Energies 2025,18, 746 19 of 77 for SoH estimation. In study [ 107 ], a pipeline optimization based on a tree and genetic algorithm is presented. Study [ 108 ] also uses a genetic algorithm as a means of parameter optimization, presenting a framework for SoH estimation, with errors of about 2%. The use of sensors for capturing battery conditions implies tabular data; however, studies were found in the portfolio, which analyze the implementation of image-based algorithms for SoH prediction. In [ 109 ], the authors propose a method capable of using only one charge and discharge cycle for SoH prediction, using image processing of current and voltage curves. Using transfer learning, the authors achieved an MAPE in the range of 10%. Transfer learning is also used in the algorithm based on battery curve image analysis in [ 110 ]. The authors analyze images of one cycle, five cycles, and ten cycles, with MAEs of about fifty, fifty-five, and sixty cycles, respectively, using eight pretrained networks, such as ResNet and GoogleNet. Other studies using algorithms from the computer vision area were found in [111–113]. Regarding the cell technologies employed, almost all the publications correspond to lithium-ion technology, among which we can highlight LFP (lithium iron phosphate), LCO (lithium cobalt oxide), NCA (lithium nickel cobalt aluminum oxide), and NMC (lithium nickel manganese cobalt oxide) battery types. Only three studies made use of battery technologies different from lithium. Study [ 114 ] analyzes the estimation of the SoHs of removed lead–acid batteries, aiming for reuse. In [ 115 ], lead–acid batteries are also analyzed, and the authors develop an SoH prediction model using LSTM networks based on charge curve data. In [ 116 ], the authors use a neural network to predict the remaining lifespan of a zinc-ion battery. 3.2. Literature Review Within the article selection process, 38 papers correspond to review papers, and they are presented in Table 4. The papers are essentially divided into reviews with qualitative analyses (e.g., trends, challenges, and general overviews), as well as papers more focused on a specific set of techniques and surveying performances and the extraction of degradation features and health indicators. In all the analyzed papers, there was no indication of the use of a methodological process for selecting the bibliographic portfolio, highlighting the importance of this work as a point of evolution within the research field. Table 4. Review papers in the bibliographic portfolio. Title Year Cited Ref. Data-driven health estimation and lifetime prediction of lithium-ion batteries: A review 2019 749 [10] Machine learning applied to electrified-vehicle-batteries’ state-of-charge and state-of-health estimation: State of the art 2020 267 [11] A review of second-life Li-ion batteries: prospects, challenges, and issues 2022 213 [12] A review of state-of-health estimations and remaining-useful-life prognostics of lithium-ion batteries 2021 200 [13] A review of non-probabilistic machine-learning-based state-of-health estimation techniques for lithium-ion batteries 2021 180 [67] Energies 2025,18, 746 20 of 77 Table 4. Cont. Title Year Cited Ref. A critical review of improved deep-learning methods for the remaining-useful-life prediction of lithium-ion batteries 2021 159 [5] Sorting, regrouping, and echelon utilization of large-scale retired lithium batteries: A critical review 2021 117 [9] Big training data for artificial-intelligence-based Li-ion diagnoses and prognoses 2020 100 [117] Machine learning in state-of-health and remaining-useful-life estimation: Theoretical and technological developments in battery degradation modeling 2022 88 [118] State-of-health prediction of lithium-ion batteries based on machine learning: Advances and perspectives 2021 81 [119] A critical review of improved deep convolutional neural networks for multi-timescale state prediction of lithium-ion batteries 2022 75 [30] A review of deep-learning approaches to predict the states of health and states of charge of lithium-ion batteries 2022 69 [26] A critical review of online battery-remaining-useful-lifetime prediction methods 2021 62 [120] Artificial neural networks, gradient boosting, and support vector machines for electric-vehicle-batteries’ state estimation: A review 2022 57 [31] State-of-health estimation and remaining-useful-life assessment of lithium-ion batteries: A comparative study 2022 43 [121] A review of modern machine-learning techniques in the prediction of the remaining useful life of lithium-ion batteries 2023 34 [122] Overview of machine-learning methods for lithium-ion-batteries’ remaining-useful-lifetime prediction 2021 33 [123] A review of machine-learning-based state-of-charge and state-of-health estimation algorithms for lithium-ion batteries 2023 33 [124] Transfer learning for batteries’ smarter-state estimation and aging prognostics: Recent progress, challenges, and prospects 2023 32 [27] Review of “gray box” lifetime modeling for lithium-ion batteries: Combining physics and data-driven methods 2022 31 [125] Energies 2025,18, 746 21 of 77 Table 4. Cont. Title Year Cited Ref. Deep-learning-enabled state-of-charge, state-of-health, and remaining-useful-life estimations for smart battery management systems: Methods, implementations, issues, and prospects 2022 26 [24] Explainability-driven model improvement for SOH estimation of lithium-ion batteries 2023 20 [126] State estimation models of lithium-ion batteries for battery management systems: Status, challenges, and future trends 2023 20 [127] State-of-charge, remaining-useful-life, and knee-point estimations based on artificial intelligence and machine learning for lithium-ion EV batteries: A comprehensive review 2022 19 [128] The development of machine-learning-based remaining-useful-life predictions for lithium-ion batteries 2023 17 [129] Comprehensive review of battery state estimation strategies using machine learning for battery management systems of aircraft propulsion batteries 2023 16 [130] A comprehensive review of lithium-ion-batteries’ state-of-health prognosis methods combining aging mechanism analysis 2023 11 [131] Research progress and application of deep learning in remaining-useful-life, state-of-health, and battery thermal management of lithium batteries 2023 11 [132] A review of the prediction of the health state and serving life of lithium-ion batteries 2022 7 [6] Specialized deep neural networks for battery health prognostics: Opportunities and challenges 2023 7 [25] Machine-learning techniques’ suitability to estimate the retained capacity in lithium-ion batteries from partial charge/discharge curves 2023 7 [133] Deep feature extraction in lifetime prognostics of lithium-ion batteries: Advances, challenges, and perspectives 2023 6 [28] Comparing deep-learning methods to predict the remaining useful life of lithium-ion batteries 2022 4 [134] Machine-learning-based remaining-useful-life prediction techniques for lithium-ion-battery management systems: A comprehensive review 2023 2 [29] Feature–target pairing in machine learning for battery health diagnosis and prognosis: A critical review 2023 2 [135] Energies 2025,18, 746 22 of 77 Table 4. Cont. Title Year Cited Ref. Research on methods for extracting aging characteristics and the health status of lithium-ion batteries based on small samples 2022 1 [136] Electric-vehicle-batteries’ capacity degradation and health estimation using machine-learning techniques: A review 2023 0 [137] Open access dataset, code library, and benchmarking deep-learning approaches for state-of-health estimations of lithium-ion batteries 2024 0 [138] Figure 12 shows that the number of review article publications within the portfolio has been increasing over the years, albeit with a considerably lower coefficient compared to that of the overall volumetric analysis. There appears to be a difference between 2022 and 2023, suggesting a potential trend toward stability in the coming years. Figure 12. Volumetric analysis of the publication year of review papers in the BP. The most cited review article in the portfolio is the study presented in [ 10 ], which examines big-data techniques regarding their feasibility and cost effectiveness in dealing with battery health in real-world applications. The methods are categorized, and advantages and limitations are identified. The authors begin by presenting methods that do not involve model training, such as the differential analysis of charge and discharge profiles, stress tests, and thermal analyses. Then, they review the use of ML for SoH estimation, highlighting the fundamental step of feature extraction. They categorize these features into three main groups: (i) model-fitted features, which depend on tests like internal resistance and are not easily accessible by sensors in a BMS (battery management system); (ii) processed external features, which are the results of differential analyses; and (iii) direct external features, which are all the variables that a sensor can collect within the battery system and can generate a large number of variables. The authors also briefly review non-probabilistic ML methods, such as artificial neural networks, SVMs, and probabilistic models, like Gaussian regression. The non-probabilistic methods are the central theme of the study presented in [ 67 ], where five types of ML algorithms for batteries’ SoH estimation are reviewed: linear regression, SVM, KNN, neural networks, and ensemble methods. The study comparatively outlines the advantages and applicability of the different methods from a theoretical standpoint. Three aspects are considered for comparing the methods: the algorithm’s Energies 2025,18, 746 23 of 77 performance based on five performance metrics (RMSE, MAE, AE, APE, and MaxE), the publication trend obtained by counting the number of publications in the last ten years, and the training modes considering feature extraction and selection. The study used 144 papers considered as relevant and published up to 10 years before (with the reference year being 2021), however, without revealing the criteria used for obtaining the portfolio. The authors conclude that neural-network-based methods and SVMs are still under research and that DL methods have shown great potential in SoH estimation under complex battery-aging conditions, especially when big data are available, and that ensemble methods, like random forest, can be considered as an emerging alternative for balancing data size and accuracy. Regarding the use of ML techniques in second-life batteries, the study presented in [ 9 ] reviewed the status and challenges of large-scale second-life applications. The authors discuss methodologies for classifying and regrouping retired batteries. They propose a rapid, multilevel, and multidimensional classification method for large-scale use. The classification method involves first solving a one-dimensional classification problem to obtain similar batteries in terms of their reaction stage. Then, a multidimensional classification is performed based on capacity and internal resistance, where usage scenarios are evaluated, for example, to determine whether the priority use is for energy or power supply. The second life is also discussed in the review presented in [ 12 ], which analyzes economic, technical, and environmental factors related to the use of second-life lithium-ion batteries, including SoH estimation methods. Regarding the reviews from this year, it is worth highlighting the study in [ 27 ], which presents the first systematic review of transfer-learning applications in the field of battery management, focusing on batteries’ state estimations and aging prognoses. The authors provide the state of the art in terms of principles, algorithmic structures, advantages, and disadvantages. For SoH estimation, a survey of papers in the field showed that transfer strategies focus on problem domain adaptation and the fine-tuning of the final model. The difficulties pointed out by the authors in using transfer learning lie in the low labeling degree of the data, which depends on the data acquisition capability at shorter intervals in a BMS. This is exacerbated by the low frequency of actual battery capacity testing during usage, especially for SoH estimation purposes. 3.3. Public Databases Analyzing the non-review papers present in the bibliographic portfolio, it was found that about 41% make use of proprietary and closed datasets, without sharing repositories for use in other studies. On the other hand, a significant and increasingly growing portion of papers conduct investigations using public datasets, comprising 59% of the non-review papers in the portfolio. As emphasized in [ 11 ], advancements in the field of ML for estimating batteries’ SoHs rely on information sharing so that new research can develop and result comparisons can occur, thereby allowing inferences about techniques that may enhance estimation accuracy. This scenario demonstrates this sharing trend, leading to faster and more voluminous developments in the research field. It is worth noting that fair comparative analyses of models/approaches also require the sharing of data splits used for training and testing/validation; only then can comparisons be made when dealing with the same population. Figure 13 demonstrates that author-provided datasets have the highest frequency of use. However, the majority of these datasets are complementary to public datasets. Among these, the highlight goes to the use of data provided by the Prognostics Center of Excellence Dataset Repository [ 139 ], from NASA, which accounts for 51% of the open datasets used in the surveyed portfolio. The dataset presented in [ 1 ], developed at the Energies 2025,18, 746 24 of 77 Massachusetts Institute of Technology (MIT), also constitutes an important data source in the surveyed papers. Figure 13. Volumetric analysis of datasets used in the papers of the bibliographic portfolio. The evolution of the proportion of closed and public datasets is presented in Figure 14. It is noticeable that the volume of applications using public datasets starts to become predominant from 2022, with the use of public datasets being about 3.2 times higher in 2023. This increase could be because of the research trend of using multiple datasets, and because more data sources are available, the application of public datasets would tend to increase. Therefore, to mitigate this effect, Figure 14 considers only the Boolean condition of whether a public dataset was used or not, and the results are similar, with the number of papers using public datasets in 2023 being about 2.3 times higher than those using closed datasets. Figure 14. Annual evolution in the BP of papers using public datasets versus closed datasets. Energies 2025,18, 746 25 of 77 Through an evaluation according to the dataset origin, Figure 15 illustrates the evolution of the dataset usage over the years in the bibliographic portfolio (BP). The increasing use of NASA datasets is noticeable, followed by the usage of the MIT [ 1 ], Oxford, and CALCE datasets. Other public datasets with even lower levels of usage are also identified in the portfolio: the Beijing Institute of Technology (BIT), Carnegie Mellon University, Stanford University, Cambridge University, the University of Hawaii, Purdue University (UL-PUR), the University of Bologna (UNIBO), and the Center for Electrochemical Energy Storage Ulm–Karlsruhe (CELEST). Figure 15. Annual evolution, in the BP, of the origin of public and author datasets. Table 5provides a summary of each of the public datasets found in the portfolio, as well as their characteristics and in which papers they were used. In total, 12 sources of public data were revealed, corresponding to 20 different datasets, all using lithium technology as the main source of the analyzed batteries. The synthesis of these databases constitutes important information for future studies, as it facilitates the selection and design of new studies on SoHs. In the NASA repository, two datasets are available for developing models aimed at estimating SoHs. The first dataset contains 34 lithium-ion 18,650 cells with a capacity of 2 Ah, undergoing processes of charging, discharging, and impedance measurements. Various temperatures are used, including 4 ◦ C, 24 ◦ C, and 44 ◦ C, with the charging process consisting of constant current until 4.2 V, followed by constant voltage until reaching the cutoff current. Different discharge regimes are adopted. The second dataset corresponds to 28 lithium-ion 18,650 cells with a capacity of 2.2 Ah that are continuously cycled with randomly generated current profiles. Reference charge and discharge cycles are also performed after a random fixed interval. In total, the cells are divided into seven equal groups, with the cycles occurring at a temperature of 40 ◦ C. In five groups, the charging cycle follows the traditional constant current–constant voltage (CC-CV) pattern, followed by randomly selected discharges. In two groups, both the charging and discharging processes are selected randomly. The cycling processes of the cells are terminated when their capacity reached either 80% or 50% of the initial capacity, depending on the type of test defined. Both NASA datasets are provided in “.mat” extension files. Energies 2025,18, 746 32 of 77 Figure 17. Frequency of groups of ML algorithms presented in the BP. Excluding the use of neural networks, tree-based methods have gained considerable representation in the portfolio. Notably, ensemble methods, such as boosting, were employed in [ 83 , 272 , 399 ], along with implementations of popular boosting algorithms, like XGBoost in [ 262 , 400 , 401 ], LightGBM in [ 402 – 404 ], and CatBoost in [ 405 , 406 ], which have gained prominence in the field of tabular data prediction in recent studies. The use of bagging can be identified in the implementation of decision-tree ensembles, such as random forest, as explored in studies [ 47 , 161 , 407 ]. Kernel-based methods, such as SVMs, can be classified as algorithms belonging to a classical and dated approach [ 408 ], yet they were considerably analyzed in the portfolio in studies [ 323 , 409 , 410 ]. The use of classical and highly interpretable linear regression was explored in 45 studies, among which, notable studies include those presented in [1,80,146,238,321,411,412]. Another approach of relative importance corresponds to algorithms that are a part of statistical methodologies, where, out of the 43 implementations in the portfolio, 37 corresponded to the use of the GPR algorithm, with examples of implementations and analyses found in [ 329 , 369 , 413 – 415 ]. As presented below, the GPR algorithm demonstrated significant usage in hybrid methodologies, ranking sixth in usage within the portfolio when considering hybridized algorithms. Another point worth noting is the interpretation conducted by studies that sought to analyze degradation through classical time-series approaches, as presented in [ 79 ], which implements the ARIMAX (AutoRegressive Integrated Moving Average Model with eXogenous input) method, and in study [ 416 ] using the ARIMA (AutoRegressive Integrated Moving Average Model) method. The NAR (Nonlinear Autoregressive) model is explored in [64]. The evolution of algorithmic categories throughout the horizon comprising the bibliographic portfolio is presented in Figure 18. It is worth noting that the authors consistently focused on exploring neural network implementations throughout the entire time horizon, with the difference from other categories maintaining a growing profile. It is possible to observe a significant increase in the implementation of decision-tree-based algorithms from 2022 to 2023. The implementation of linear models has also been gaining momentum, mainly because of the comparisons that these simpler models can offer compared to more complex algorithms. Additionally, they present greater interpretability of variables and, therefore, of the modeling [ 6 , 104 , 417 ]. The use of time-series techniques has remained relatively constant in the portfolio, which, in contrast to the increasing volume of publications per year, indicates that the percentage of implementation compared to that of other Energies 2025,18, 746 33 of 77 categories has been decreasing. Algorithms related to clustering, quantile regression, the neighborhood method, and unsupervised learning were more recently implemented within the portfolio, between 2022 and 2023. Figure 18. Evolution of algorithmic implementation in the BP by category. Going deeper into the analysis of the two main categories of algorithms implemented in the portfolio, the graphs in Figure 19 depict the distributions of neural network and decision-tree algorithms. In the neural network category, following the observations from the overall analysis, there is a dominance of LSTM and CNN networks, followed by simple neural networks, algorithms based on well-known networks, such as extreme-learning machines and RNNs. In the decision-tree algorithms, there is a predominance of ensemble bagging using the random forest algorithm, accounting for 35% of the tree implementations in the portfolio, followed by boosting algorithms, such as XGBoost, GBT, LGBM, and Adaboost. A detailed exploration of this type of ensemble can be observed in battery degradation studies, with approximately 60% of the tree implementations in the portfolio. (a) (b) Figure 19. Frequencies of implemented algorithms: (a) decision trees; (b) neural networks. When analyzing the evolution of techniques implemented in the portfolio, as depicted in Figure 20, it is noticeable that the use of LSTM networks predominates throughout almost the entire analyzed time horizon. The use of CNN networks began to gain prominence from publications in 2021. The use of the random forest became the third most implemented technique in the portfolio’s works in 2023; however, the use of simple ANNs showed a sharp decline in the last year. Implementations of the SVM method seem to be decelerating, with a decrease in usage in 2021 and maintaining the number of implementations in 2023 Energies 2025,18, 746 34 of 77 compared to 2022. The use of GRU networks also appears to be trending, becoming the fourth most implemented algorithm in 2023. As a baseline and comparative algorithm, linear regression also demonstrates an increase in the number of implementations over the horizon. Other algorithms that seem to be experiencing a growing exploration are DNN, GPR, and XGBoost. Figure 20. Evolution of algorithmic implementation in the bibliographic portfolio. The evolution of the portfolio’s main neural network implementations is presented in the graph in Figure 21. As previously highlighted, the use of LSTM and CNN networks is at the forefront of authors’ research in the field, with LSTM networks being the main technique in this category from 2020 onward, and CNNs gaining prominence from 2021. It is noteworthy to highlight some recent jumps in implementations in the portfolio, from 2022 to 2023, such as the exploration of GRU, DNN, ELM, and BPNN techniques. It is striking to see the resurgence of the exploration of more classical networks, such as BPNNs, by authors in the field. Figure 21. Evolution of neural network algorithmic implementation in the BP. Because of its secondary prominence in the portfolio, we also present the evolution of decision-tree-based algorithms in Figure 22. The evolution of the random forest algorithm’s usage over the years can be observed, with a notable increase in 2023, and the possible replacement of GBT boosting by newer versions, such as XGBoost and LightGBM. Energies 2025,18, 746 35 of 77 Figure 22. Evolution of decision-tree-based algorithmic implementation in the portfolio. 3.4.1. Deep-Learning Models ML, as a subfield of artificial intelligence, employs algorithms and statistical techniques to construct predictive models. Neural networks represent a subset of ML algorithms that have seen their structural complexity increase over time, in tandem with computational advancements. This complexity primarily manifests in the augmentation of intermediate layers within networks, enhancing the algorithm’s ability to discern patterns and giving rise to a subfield known as DL algorithms [ 418 , 419 ]. Traditional ML algorithms often outperform DL methods in scenarios of limited data availability. However, as datasets expand, traditional ML algorithms tend to reach performance plateaus, while DL algorithms demonstrate significant superiority over other learning strategies [418]. The expected potential of DL techniques can be observed in the bibliographic portfolio. Within the set of techniques belonging to neural-network-based algorithms, 307 papers using DL algorithms were identified, representing a significant 57.5% of the portfolio. This demonstrates a strong trend within this research field. DL algorithms were considered as those with more than three hidden layers. Although there is no consensus among authors and researchers in the field regarding the exact number of layers required to characterize a network as DL, some references consider this number of layers to indicate “light” DL networks, while “heavy” DL networks can have from tens to hundreds of hidden layers [420,421]. A comparative analysis of the evolution of the proportion of DL usage in neural network algorithms is shown in Figure 23. There is noticeable stability in the proportion, discounting the factor of publication volume in the early years, which settles between 80% and 90% in the last 3 years of the portfolio. Energies 2025,18, 746 36 of 77 Figure 23. Proportion of DL implementation in neural network techniques in the portfolio. The volume of DL technique implementations in the portfolio is presented in Table 7, with a visual proportion overview shown in Figure 24. In Table 7, the term “Frequency” refers to the number of implementations recorded for each technique. Papers in the portfolio may present more than one implementation within the same study. Together, LSTM and CNN techniques account for 60% of the implementations, while RNN, DNN, and GRU techniques stand out, with implementations in more than 20 papers each. In total, 23 techniques were implemented only once. Table 7. Survey of DL techniques implemented by authors in the bibliographic portfolio. Algorithm Frequency Algorithm Frequency Algorithm Frequency LSTM 161 RESNET 2 PKNN 1 CNN 86 ELM 2 DBNN 1 RNN 29 BPNN 2 DSMTNET 1 DNN 27 EFFICIENTNET 1 DCN 1 GRU 26 CRNN 1 DDAN 1 ANN 10 VISION TRANSFORMER NETWORK 1 DEEP REINFORCEMENT LEARNING 1 MLP 8 VGG11 1 DELM 1 TCN 6 TRANSFORMER NEURAL NETWORK 1 GOOGLENET 1 ENN 5 TDNN 1 DENSENET 1 FFNN 5 BNN 1 ALEXNET 1 GRAPH NN 4 CAPSNET 1 EDFM 1 DBN 3 REGRESSIVE MATCHING NETWORK 1 DILATED RESIDUAL NETWORK 1 DCNN 3 CDTSGANN 1 FCNN 1 Energies 2025,18, 746 37 of 77 Figure 24. Distribution of techniques in DL implementations in the portfolio. The evolution of the main DL algorithms implemented in the portfolio is presented in Figure 25, which shows that the most implemented algorithm is the LSTM network. In addition to the rising trends of LSTM and CNN networks, we again highlight the recent implementation trends of DNN and GRU algorithms, as well as the first relevant implementations of the DCNN algorithm, found in 2023, which combines characteristics of DNN and CNN networks. Figure 25. Evolution of DL algorithmic implementation in the BP. In [ 44 ], which is the main DL publication according to the citation count, the authors employ a hybrid LSTM-RNN model to capture long-term information regarding the relationship between a battery’s capacity and its degradation, emphasizing that such a dual approach is recommended to avoid overfitting issues. Another work utilizing a hybrid technique based on LSTM is presented in [ 48 ], using an Elman neural network (ENN). The concept of transfer learning, which involves the use of neural networks trained and fine-tuned in large datasets and then fine-tuned on their final layers in a specific dataset to transfer knowledge to another problem, is discussed in [ 65 ], which implements an LSTM network. Other examples of studies using LSTMs can be found in [51,59,422,423]. Regarding relevance by citation count, the main studies using DL are presented in Table 8, where it is notable that the use of LSTM is present in six out of the ten studies. Another interesting point is the use of a hybrid approach by the publications, using two DL algorithms in this case. The table also highlights the datasets used by the authors, with half the publications utilizing public data Additionally, the table includes a marking to Energies 2025,18, 746 38 of 77 indicate whether the implementation was hybrid, where more than one algorithm was used to determine the same prediction. Table 8. Main publications in the portfolio with DL implementation. Algorithm Hybrid Dataset Title Year Cited Ref. LSTM, RNN Yes Author Long short-term memory recurrent neural network for remaining-useful-life prediction of lithium-ion batteries 2018 880 [44] LSTM, GPR Yes Author A data-driven approach with uncertainty quantification for predicting future capacities and remaining useful life of lithium-ion batteries 2021 434 [45] LSTM, ENN Yes Author Remaining-useful-life prediction for lithium-ion batteries based on a hybrid model combining the long short-term memory and Elman neural networks 2019 316 [48] DNN No NASA Remaining-useful-life prediction for lithium-ion batteries: A deep-learning approach 2018 313 [49] CNN, LSTM Yes NASA A data-driven auto-CNN-LSTM prediction model for lithium-ion-batteries’ remaining useful life 2021 291 [50] LSTM No NASA State-of-health estimation and remaining-useful-life prediction for lithium-ion batteries based on a variant long short-term memory neural network 2020 284 [51] DCNN No Author A deep-learning method for online capacity estimation of lithium-ion batteries 2019 260 [53] DNN No CALCE, NASA, MIT, OXFORD Machine-learning pipeline for batteries’ state-of-health estimations 2021 246 [54] LSTM No NASA A neural-network-based method for RUL prediction and SOH monitoring of lithium-ion batteries 2019 245 [55] PKNN No Author A novel estimation method for the states of health of lithium-ion batteries using a prior-knowledge-based neural network and a Markov chain 2019 239 [56] The five most recent studies with DL implementation in the bibliographic portfolio are presented in Table 9. The publications correspond to the year 2024, which, in total, had 17 publications on the subject in the first two weeks of the year (the total number of portfolio publications in 2024 was 21). The five highlighted papers make use of public datasets (16 out of 17 in total for the year), with two publications implementing a hybrid approach with DL (seven out of seventeen in total for the year), including the use of a decision-tree-based algorithm. Energies 2025,18, 746 39 of 77 Table 9. Recent publications in the portfolio with DL implementation. Algorithm Hybrid Dataset Title Year Cited Ref. GCN No NASA, OXFORD State-of-health and remaining-useful-life predictions of lithium-ion batteries with a conditional graph convolutional network 2024 2 [179] RNN No MIT Jellyfish-optimized recurrent neural network for state-of-health estimations of lithium-ion batteries 2024 2 [336] LSTM No NASA, CALCE Remaining-useful-life predictions of lithium Batteries based on a CNN–Mogrifier LSTM-MMD 2024 1 [192] MLP, GRU Yes NASA, CALCE An MLP–mixer and mixture of expert models for remaining-useful-life predictions of lithium-ion batteries 2024 0 [220] RF, GRU Yes NASA State-of-health estimations for lithium-ion batteries using a random forest and a gated recurrent unit 2024 0 [221] 3.4.2. Hybrid Models Hybrid ML models combine different ML techniques and algorithms to enhance prediction performance by leveraging the strengths of each method while compensating for their individual weaknesses [ 48 , 424 ]. Within the bibliographic portfolio, a total of 135 publications have implemented this approach. The evolution of hybrid model usage in the portfolio is depicted in Figure 26. As shown, it can be inferred that the use of hybrid approaches in SoH estimation is a recent field of exploration, with a significant increase in implementations in 2023. Considering the volume of portfolio publications, the use of hybrid approaches represents approximately 30% of the papers surveyed, a jump of nearly 50% compared to 2022, where it was present in about 18% of publications. In the first two weeks of 2024, a total of nine papers with hybrid approaches were published. Figure 26. Evolution of hybrid algorithmic implementation in the BP. Energies 2025,18, 746 40 of 77 The most utilized techniques in the hybrid modeling approach also correspond to the use of DL networks, such as LSTM and CNN, with a considerable advantage, with 74 and 51 implementations, respectively, as indicated in Figure 27. Other DL algorithms, such as GRU and RNN, are also notable, for example, in [ 232 , 233 , 381 , 425 ]. Other classical algorithms, such as SVM and GPR, found in [ 149 , 426 ], and decision-tree-based algorithms, like RF, XGBoost, and LightGBM, present in [80,221,427], are also noteworthy. Figure 27. Frequency of techniques addressed in papers with hybrid algorithms in the BP. The evolution of the implementations of the main algorithms is presented in Figure 28. It is possible to observe the increases in the implementations of LSTM and CNN networks, in line with previous results, and the recent evolution of the use of GRU and RF algorithms, with a peak in usage in 2023. Figure 28. Evolution of hybrid algorithmic implementation in the bibliographic portfolio. Energies 2025,18, 746 41 of 77 Figure 29 presents the found combinations resulting from the analysis of hybrid algorithms in the portfolio. The primary combination occurs with the LSTM and CNN networks, with 27 implementations in the portfolio. Algorithms that appear individually in the survey reflect either a hybrid approach (e.g., the integration of different configurations of the same algorithm, such as combining a 2-dimensional CNN with a 3-dimensional CNN), or methodologies that incorporate filters (e.g., the Kalman filter) and optimization algorithms as a part of their design. In total, 65 papers presented combinations of algorithms that were implemented only once in the portfolio, indicating that many researchers still evaluate different approaches of hybrid models. Figure 29. Combinations of algorithms found in papers with a hybrid approach in the BP. The combinations of each algorithm in the portfolio are presented in Table 10, allowing for the identification of hybrid approaches, evaluated by the authors, within the portfolio, which can serve as a starting point for testing hybrid models in new research. A visualization of these combinations is shown in Figure 30, where centers of algorithmic connections can be observed, revolving around LSTM, CNN, RF, and GPR techniques. As demonstrated in Table 10 and Figure 30, LSTM networks exhibit a considerable range of combinations with other algorithms, including decision-tree, statistical, kernel, neighborhood, and clustering algorithms, as well as other neural network architectures. Table 10. Connections between algorithms in papers with a hybrid approach in the BP. Algorithm Algorithmic Connections LSTM DCNN, FFNN, CNN, ANN, ENN, DNN, TCN, RNN, XGBOOST, BMA, GRAPH NEURAL NETWORK, SVM, GPR, DBN, GRU, RANDOM FOREST, FUZZY CLUSTERING, BPNN, LINEAR QUANTILE REGRESSION, MLP, RESNET, ADABOOST RANDOM FOREST ANN, NAR, LINEAR REGRESSION, GRADIENT-BOOSTING DECISION TREE, GPR, LIGHTGBM, XGBOOST, LSTM, SVM, RBFNN, RIDGE REGRESSION, KNN, GRU, EXTRATREES, ELM SVM ARIMA, DECISION TREE, ELM, LSTM, GPR, RBFNN, RANDOM FOREST, RIDGE REGRESSION, LINEAR REGRESSION, GRU, RNN Energies 2025,18, 746 48 of 77 difficult. In these cases, the comparisons are empirical, and there is an associated probability of a particular approach being better than another. Among the arguments used are concerns related to the quality of the data from some cells, as well as supposed bias in the training split based on differences in the distribution of the cycles used in [ 1 ]. Another point that drew attention was cases where authors performed splits of training, testing, and validation while keeping data samples from all the cells in each set, which impacts the reliability of the results presented, as per the performances in the articles highlighted below. Table 15 summarizes the RUL (remaining-useful-life) prediction performances in publications that had the same validation set, totaling five papers. The validation sets are referred to as the 1 ◦ Test and 2 ◦ Test by the authors in [ 1 ]. The 1 ◦ Test includes batteries under the same cycling conditions as those of the training set, while the 2 ◦ Test corresponds to batteries with a different usage profile. Notably, the performance gain achieved in [ 335 ] is highlighted, where the MAPE errors in both test sets are reduced by over 50% compared to that of the baseline study [ 1 ]. This improvement was achieved using the same 100 cycles of information for the prediction and implementing a hybrid model using the LSTM DL technique along with the GPR algorithm. However, such significant results were not found in the use of the LSTM-CNN combination in [ 333 ], where a range of errors similar to that of the baseline study [ 1 ] was observed despite employing more complex techniques. In [ 317 ], an increase in performance is evident with a hybrid approach involving neural networks, linear regression, and RF, using a reduced set of 80 cycles. These findings suggest that efforts in algorithmic selection do not necessarily guarantee higher performance, and steps such as feature construction and selection may represent an even more relevant stage in research. In Table 16, the rest of the performance survey with the MIT dataset, conducted in the bibliographic portfolio, is presented. Here, the authors did not maintain the same modeling and validation splits, and there are variations in the target variables. Therefore, all the comparisons made may exhibit significant bias. The targets described in the table are presented to maintain the nomenclature adopted by the authors. The target’s “early battery lifetime”, also referred to in studies as the “early cycle life”, aims to determine the total number of cycles a battery will present based on data from the first cycles of a battery. The “end of life” in the analyzed studies is related to determining the total number of cycles considering data from the last cycles, without necessarily knowing the entire battery history. Therefore, it is generally accompanied by models that determine the remaining number of cycles (RULs) and/or the current cycle. The “capacity” target was linked to studies that used regression models in predicting time series (“capacity trajectories”), where the evolution of the battery capacity over time is obtained or the prediction on a short horizon, such as the discharge capacity in the next cycle, can be used for SoH updates. Energies 2025,18, 746 49 of 77 Table 15. Prediction performance of studies using the same samples for validation with MIT dataset. Algorithm 1◦Test 2◦Test Cycles Title Year Cited Ref. RMSE MAPE RMSE MAPE Linear regression 118 14.1 214 10.7 100 Data-driven prediction of the battery cycle life before capacity degradation 2019 811 [1] RF, linear regression, ANN 80 9.8 174 7.5 80 Prognostics of the battery cycle life in the early-cycle stage based on a hybrid model 2021 41 [317] Ridge Reg 125 - 188 -100 Statistical learning for accurate and interpretable battery lifetime predictions 2021 30 [102] Enet Reg 132 196 RF 141 197 MLP 140 218 CNN 72 204 CNN, MLP 114 8.54 178 11.31 100 A hybrid ensemble deep-learning approach for the early prediction of batteries’ remaining useful life 2023 9 [333] GPR, LSTM 30 5.52 51 5.35 100 Joint modeling for early predictions of Li-ion-batteries’ cycle life and degradation trajectory 2023 3 [335] Table 16. Prediction performance of studies in the portfolio using different samples for validation with the MIT dataset. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year Bayesian ridge Capacity (Ah) 0.45 0.76 - Predicting capacity considering only a short portion of partial charge/discharge data - Requires a 15 min sample of operation - Utilizes charging and discharging steps - 63 cells for training, 10 for calibration, 51 for testing (split based on the distribution of cycle numbers in the dataset, maintaining the same distribution across all the sets) [54]2021 GPR 1.00 1.91 RF 0.11 0.14 DNN 0.23 0.45 CNN RUL 10.6 76 - Utilization of four cycles - Incorporation of charging and discharging steps - 86 cells for training, 19 for validation, 19 for testing [8] 2020 Energies 2025,18, 746 50 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year GBT RUL 7.5 84.9 58.6 0.94 - Usage of 250 cycles - Incorporation of the discharge stage - Data split into 2/3 for training and 1/3 for testing; no specification if cells from the training set are excluded from the test set; the splitting process is repeated four times, and the performances are analyzed - Performance corresponds to the average of the four cases [83] 2020 CNN Early battery lifetime 3.80 (1) 42 (1) 33 (1) - Testing for the use of the first 20 (1), 40 (2), 60 (3), 80 (4), 100 (5) cycles for battery life prediction - Utilization of the first five cycles and the last fifteen cycles for RUL prediction - Incorporation of the charging stage - 94 cells for training, 30 for testing [315]2021 1.30 (2) 19 (2) 13 (2) 1.12 (3) 13 (3) 11 (3) 1.21 (4) 13 (4) 10 (4) 1.12 (5) 11 (5) 9 (5) RUL 3.55 11 9 DNN End of life 7.78 (1) 57 (1) - Testing for the use of the last 1 (1) to 100 (2) cycles. - Incorporation of the discharging stage - EoL = current cycle + cycles used for data collection + RUL. - 65 cells for training, 16 for testing (discarding 43 cells) - Majority of RMSE for RUL < 50 cycles, larger errors for cells with fewer than 100 cycles - * Errors for predicting the current life cycle increase infinitely for cells with over 750 cycles (author’s justification based on the low sample quantity for this scenario) [318]2022 3.97 (2) 33 (2) Cycle life <65 (1) <40 (2) >90 * RUL <65 (1) <40 (2) Energies 2025,18, 746 51 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year SVR Capacity trajectory (Ah) 1.61 3.22 - Use of the last 20 cycles - Estimates the evolution of capacity trajectory over time until EoL (time series using regression) - Incorporation of both charging and discharging stages. - 84 cells for training, 40 for testing [253]2022 RF 0.93 2.12 GPR 1.35 2.58 ANN 1.13 1.92 CNN RUL 4.15 27.47 16.09 - Use of 10 cycles - Incorporation of the charging stage - 70% of the data for training, 30% for testing (does not specify if cells from the training set were excluded from testing) [319] 2022 Linear reg, (1) RUL 90 53.81 * - Use of 10 cycles - Does not exclude cells from training during testing; 60% of the data for training, 20% for validation, and 20% for testing - Incorporation of the charging stage - Classification model to predict if a battery has less than 150 cycles of RUL or 150 cycles or more of RUL - RUL Approaches: - (1): Does not consider the classification model - (2): Regression model for each predicted RUL class in the classification model - * Considering cases where RUL > 150 cycles: 18.51, 10.51, 9.79, respectively - For capacity, the author evaluated 100 **, 150 ***, and 200 **** cycles ahead [321]2021 MLP (1) 52 23.03 * Logistic reg. + MLP (2) 49 15.2 * MLP Discharge capacity after “x” cycles. 0.24 ** 0.45 *** 0.64 **** Energies 2025,18, 746 52 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year Transfer Learning (CNN + RNN + “fully connected”) Capacity (Ah) 0.176 * 2.57 * 0.999 * - Use of the last 30 cycles to predict the capacity of the next cycle and RUL - The author does not assess the performance of estimating the capacity trajectory for horizons longer than one cycle - Use of the charging stage - Use of the MIT dataset to train a model and evaluate the performance of the model with transfer learning on a dataset constructed by the author - Author’s dataset contains information from 77 LFP/graphite cells of 1.1 Ah. - 22 cells separated for testing - * Performance considering training with the author’s dataset. - ** Performance considering transfer learning from a model pretrained with the MIT dataset [385]2022 0.328 ** 4.65 ** 0.997 ** RUL 8.72 * 186 * 0.804 * 9.80 ** 240 ** 0.770 ** Elastic net RUL 5.21 43.38 0.98 - Use of 100 cycles - Use of both charging and discharging stages - No exclusion of cells from training in testing; 70% of the data for training, 30% for testing [329]2022 GPR 5.26 43.71 0.98 SVM 5.88 53.04 0.97 RF 8.17 84.69 0.92 DT ensemble 7.93 88.74 0.91 XGBoost 7.92 91.13 0.92 RVM 10.32 96.21 0.89 DT 9.59 106.62 0.87 Energies 2025,18, 746 53 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year CNN, LSTM Cycle life 2.28 (1) 19 (1) 14 (1) 0.9980 (1) - Testing of usage from the last 50 (1), 60 (2), 70 (3), 80 (4), 90 (5), 100 (6) cycles - Usage of both charging and discharging stages - 93 cells for training and 31 for testing (split based on the distribution of cycle numbers in the dataset, maintaining the same distribution in all the sets) [327]2022 4.59 (2) 50 (2) 33 (2) 0.9869 (2) 3.02 (3) 25 (3) 18 (3) 0.9967 (3) 3.43 (4) 25 (4) 19 (4) 0.9967 (4) 1.84 (5) 16 (5) 13 (5) 0.9985 (5) 1.47 (6) 11 (6) 9 (6) 0.9993 (6) RUL 2.16 (1) 12 (1) 8 (1) 0.9993 (1) 3.17 (2) 15 (2) 12 (2) 0.9989 (2) 1.93 (3) 11 (3) 8 (3) 0.9994 (3) 1.85 (4) 14 (4) 10 (4) 0.9990 (4) 1.72 (5) 13 (5) 9 (5) 0.9992 (5) 1.25 (6) 8 (6) 6 (6) 0.9997 (6) Graph Neural Network Capacity trajectory (Ah) 0.009 * 0.0377 * 0.9399 * - Using 350 measurement points as input. - Usage of the charging stage. - 70% of the cells used for training and 30% for testing. - Estimates the evolution of capacity trajectory over time until End of Life (time series using regression). - * Performance based on the worst predicted cell. - ** Performance based on the best predicted cell. [92]2023 0.004 ** 0.0025 ** 0.9894 ** Energies 2025,18, 746 54 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year LightGBM SoH (%) 1.751 - Estimation of SoH based on 300 s measurements - Usage of the discharge stage - Cells 91 and 100 from the MIT dataset are used for training, cell 124 used for testing - * Model considering LightGBM, XGBoost, Random Forest (RF), SVR, GPR as base models, and linear regression as a meta-model [99]2023 XGBoost 1.616 RF 1.721 SVR 1.926 GPR 1.539 Stacking 1.489 * LSTM SoH (%) after “x” cycles 0.016 (1) 1.81 (1) 0.0098 (1) - Testing prediction horizons of 25 (1), 50 (2), 100 (3), 150 (4), 200 (5), 250 (6), 300 (7), 350 (8), 400 (9) cycles ahead - Usage of charge and discharge stages - 64% of the cells are used for training, 20% for validation, and 16% for testing (cells cannot be in more than one set) - Average of performances per cell [144]2023 0.021 (2) 2.30 (2) 0.0130 (2) 0.024 (3) 2.80 (3) 0.0140 (3) 0.024 (4) 2.86 (4) 0.0120 (4) 0.031 (5) 3.60 (5) 0.0180 (5) 0.026 (6) 3.00 (6) 0.0150 (6) 0.030 (7) 3.49 (7) 0.0200 (7) 0.032 (8) 3.70 (8) 0.0200 (8) 0.033 (9) 3.80 (9) 0.0201 (9) Energies 2025,18, 746 55 of 77 Table 16. Cont. Algorithm Target MAPE RMSE RMSPE MAE MRE R2Observations Ref. Year RF Cycle life 0.57 4.65 - Usage of 100 cycles. - Utilization of charge and discharge stages. - 75% of the data used for training, 25% used for testing (cells from the training set were not excluded from testing). - Discrepant results compared to the literature, potential model validation error by the author. [334] 2022 ResNet50 Early lifetime 119.98 0.8501 - Use of the first 100 cycles for predicting the total lifespan. - Utilization of images obtained from plots with voltage and capacity information as features. - 80 cells for training and 43 cells for testing, process repeated five times. [111]2024 CNN 115.85 0.8557 LeNet 129.77 0.8197 AlexNet 91.51 0.9121 VGG16 122.19 0.8466 Energies 2025,18, 746 56 of 77 Through a comparison analysis, it is possible to observe performance improvements compared to the baseline study in [ 315 ], with a MAPE of around 3.5% using a CNN. In [ 8 ], the authors achieved significant performance with a CNN and only four cycles of data as input to the algorithm, a reduction that provides new perspectives for the use and conditioning of batteries. Using TL, the authors in [ 385 ] achieved results within the error magnitudes of the studies that use only one dataset, demonstrating that TL can be a useful tool for aggregating the volume of information from other types of experimental tests and cell technologies to overcome data limitations. In [ 327 ], the best results for RUL prediction were achieved, with MAPE values below 2%. The authors conducted tests considering different data usage intervals, ranging from 50 to 100 cycles, and using a hybrid LSTMCNN approach. Conversely, in [ 334 ], the authors claim errors around half a cycle, well below those presented in various consulted studies. The use of data from all the cells during training ends up bringing a possible leakage when validating the algorithm because the pattern of all the cells was passed to the model, which, consequently, did not develop proper learning but possible “rote memorization”, associating levels of variable values with related life cycles. This point highlights the importance of correctly analyzing the results for the dissemination of research in the field. This analysis reveals significant challenges in comparing RUL prediction models because of inconsistencies in data splitting, target variables, and evaluation metrics across different studies. Although advancements have been observed, particularly with hybrid models, like LSTM-GPR and the application of TL, the lack of standardized methodologies hinders direct comparisons and hinders the identification of truly superior approaches. The use of limited data cycles in training, as demonstrated in [ 8 ], and the exploration of feature engineering, as suggested by the results in [ 317 ], present promising avenues for future research. However, it is crucial to emphasize the importance of rigorous data splitting procedures, avoiding data leakage, as observed in [ 334 ], to ensure the reliability and generalizability of the obtained results. Notably, for most of the analyzed models, MAPE errors of around 10% have become achievable with the development of algorithms and open datasets. Moving forward, establishing standardized datasets and evaluation protocols will be essential to facilitate progress in the field and enable more meaningful comparisons between different RUL prediction models. 3.6. The Importance of SoH in Smart Systems, Energy Informatics, and Smart Grids Accurate battery SoH estimates, derived from ML algorithms and analyses based on large datasets, have significant implications for energy informatics and intelligent systems, such as smart grids. This study explores some of the key applications connecting SoH prediction to improvements in energy efficiency and sustainability. Energy Informatics and Energy Management in Smart Grids: Energy informatics, the integration of information systems and energy, plays a fundamental role in the efficient management of smart grids. Accurate SoH estimation enables more effective management of second-life batteries by integrating them into storage and distribution networks. This approach not only reduces waste but also enhances the reliability and resilience of electrical grids, especially in contexts involving renewable energy sources [1,10,12]. IoT Devices and Sustainability: SoH prediction models based on DL techniques, such as LSTM and CNN networks, facilitate the preventive maintenance of IoT devices that rely on batteries. These models support a more sustainable economy by optimizing replacement cycles and extending the lifespans of connected smart devices [ 48 , 54 , 64 ]. The use of these devices in smart grids also reduces reliance on manual interventions, promoting greater automation and efficiency [9,12]. Energies 2025,18, 746 57 of 77 Real-Time Monitoring and Control: Wireless sensor networks (WSNs) integrated with SoH algorithms offer real-time monitoring capabilities, essential for dynamic system adjustments. In smart grids, this enables load balancing and the optimization of the energy distribution, improving the overall system performance [54]. Environmental Impact and Sustainability: The reuse of batteries, underpinned by reliable SoH estimates, contributes to a circular economy by reducing the environmental impact and the carbon footprint associated with the production of new batteries [ 9 , 12 ]. Hybrid models, such as the combination of LSTM with Elman neural networks (ENNs), have already demonstrated accuracy gains of up to 50% compared to classical approaches, increasing confidence in battery reuse for storage systems and smart grids [48]. Through these applications, SoH prediction not only enhances the management and efficiency of smart grids but also reinforces the connections among energy informatics, sustainability, and technological innovation. This highlights the importance of robust prediction methods for the future of intelligent energy systems. Additionally, it is essential to emphasize that accurate SoH prediction significantly contributes to the evolution of intelligent systems by reducing operational uncertainties and enabling the seamless integration of emerging technologies. The precise forecasting of SoH enhances system reliability by enabling the optimized allocation of energy resources, such as second-life batteries, across diverse use cases. These advancements also support the adoption of predictive maintenance systems, which reduce operational costs while maximizing energy efficiency and long-term sustainability. By converging machine-learning techniques, such as deep neural networks, with advanced data management platforms, SoH becomes a critical metric for decision-making in smart grids and the IoT, driving resilience and sustainability in energy infrastructure. 4. Conclusions This study highlights the growing importance of ML techniques in estimating the SoHs of batteries, as evidenced by a systematic bibliographic portfolio analysis. The application of ProKnow-C enabled the objective selection of 534 relevant papers from an initial pool of 6032 publications, providing a structured and replicable methodology for characterizing research within this domain. The results reveal several key trends. First, there has been a significant increase in scientific production in this area, particularly since 2022, with 40% of the selected papers published in 2023. The increasing relevance of battery reuse, driven by the expansion of the electric vehicle market, is expected to further boost research in SoH estimation. Second, the analysis highlights the importance of open datasets, with 60% of the reviewed studies using publicly available data. The NASA Prognostics Center of Excellence repository remains the most cited source, accounting for over half of the open data usage. Overall, the portfolio analysis revealed the presence of 12 available open data sources, with 6 of these sources published in the years 2022 and 2023. From a methodological perspective, DL techniques, especially LSTM networks and CNNs, dominate the field, with DL accounting for 58% of the implementations. Hybrid approaches, including those combining LSTM and CNNs, are increasingly prominent, representing approximately 25% of the reviewed studies. The emergence of TL in publications since 2022 also highlights a promising avenue for leveraging diverse datasets to address data scarcity and heterogeneity in SoH modeling. Performance evaluations based on the MIT dataset indicate that classical approaches achieve mean absolute percentage errors of approximately 10%, whereas DL techniques have reduced errors by 50% in some cases. Some studies report prediction errors as low as 1–4% using CNNs, emphasizing the potential of advanced algorithms in this field. Energies 2025,18, 746 64 of 77 80. Li, F.; Zhang, L.; Chen, B.; Gao, D.; Cheng, Y.; Zhang, X.; Yang, Y.; Gao, K.; Huang, Z. An Optimal Stacking Ensemble for Remaining Useful Life Estimation of Systems Under Multi-Operating Conditions. IEEE Access 2020,8, 31854–31868. [CrossRef] 81. Liu, G.; Zhang, X.; Liu, Z. State of Health Estimation of Power Batteries Based on Multi-Feature Fusion Models Using Stacking Algorithm. Energy 2022,259, 124851. [CrossRef] 82. Paulson, N.H.; Kubal, J.; Ward, L.; Saxena, S.; Lu, W.; Babinec, S.J. Feature Engineering for Machine Learning Enabled Early Prediction of Battery Lifetime. J. Power Sources 2022,527, 231127. [CrossRef] 83. Yang, F.; Wang, D.; Xu, F.; Huang, Z.; Tsui, K.-L. Lifespan Prediction of Lithium-Ion Batteries Based on Various Extracted Features and Gradient Boosting Regression Tree Model. J. Power Sources 2020,476, 228654. [CrossRef] 84. Greenbank, S.; Howey, D.A. Piecewise-Linear Modelling with Automated Feature Selection for Li-Ion Battery End-of-Life Prognosis. Mech. Syst. Signal Process. 2023,184, 109612. [CrossRef] 85. Wu, J.; Su, H.; Meng, J.; Lin, M. State of Health Estimation for Lithium-Ion Battery via Recursive Feature Elimination on Partial Charging Curves. IEEE J. Emerg. Sel. Top. Power Electron. 2023,11, 131–142. [CrossRef] 86. Greenbank, S.; Howey, D. Automated Feature Extraction and Selection for Data-Driven Models of Rapid Battery Capacity Fade and End of Life. IEEE Trans. Ind. Inform. 2022,18, 2965–2973. [CrossRef] 87. Peng, J.; Zheng, Z.; Zhang, X.; Deng, K.; Gao, K.; Li, H.; Chen, B.; Yang, Y.; Huang, Z. A Data-Driven Method with Feature Enhancement and Adaptive Optimization for Lithium-Ion Battery Remaining Useful Life Prediction. Energies 2020,13, 752. [CrossRef] 88. Ren, Z.; Du, C.; Ren, W. State of Health Estimation of Lithium-Ion Batteries Using a Multi-Feature-Extraction Strategy and PSO-NARXNN. Batteries 2022,9, 7. [CrossRef] 89. Fei, Z.; Zhang, Z.; Yang, F.; Tsui, K.-L.; Li, L. Early-Stage Lifetime Prediction for Lithium-Ion Batteries: A Deep Learning Framework Jointly Considering Machine-Learned and Handcrafted Data Features. J. Energy Storage 2022,52, 104936. [CrossRef] 90. Tang, T.; Yuan, H. The Capacity Prediction of Li-Ion Batteries Based on a New Feature Extraction Technique and an Improved Extreme Learning Machine Algorithm. J. Power Sources 2021,514, 230572. [CrossRef] 91. Wang, Z.; Liu, N.; Guo, Y. Adaptive Sliding Window LSTM NN Based RUL Prediction for Lithium-Ion Batteries Integrating LTSA Feature Reconstruction. Neurocomputing 2021,466, 178–189. [CrossRef] 92. Wang, Z.; Yang, F.; Xu, Q.; Wang, Y.; Yan, H.; Xie, M. Capacity Estimation of Lithium-Ion Batteries Based on Data Aggregation and Feature Fusion via Graph Neural Network. Appl. Energy 2023,336, 120808. [CrossRef] 93. Cui, Z.; Gao, X.; Mao, J.; Wang, C. Remaining Capacity Prediction of Lithium-Ion Battery Based on the Feature Transformation Process Neural Network. Expert Syst. Appl. 2022,190, 116075. [CrossRef] 94. Yao, X.-Y.; Chen, G.; Hu, L.; Pecht, M. A Multi-Model Feature Fusion Model for Lithium-Ion Battery State of Health Prediction. J. Energy Storage 2022,56, 106051. [CrossRef] 95. Fu, P.; Chu, L.; Hou, Z.; Guo, Z.; Lin, Y.; Hu, J. State-of-Health Prediction Using Transfer Learning and a Multi-Feature Fusion Model. Sensors 2022,22, 8530. [CrossRef] 96. Himadri Sekhar, B.; Sindhu, S.; Amalendu Bikash, C.; Chandan Kumar, C. State-of-Health Estimation and End of Life Prediction for the Lithium-Ion Battery by Correlatable Feature-Based Machine Learning Approach. Int. J. Perform. Eng. 2021,17, 825. [CrossRef] 97. Luo, C.; Zhang, Z.; Qiao, D.; Lai, X.; Li, Y.; Wang, S. Life Prediction under Charging Process of Lithium-Ion Batteries Based on AutoML. Energies 2022,15, 4594. [CrossRef] 98. Lee, G.; Kim, J.; Lee, C. State-of-Health Estimation of Li-Ion Batteries in the Early Phases of Qualification Tests: An Interpretable Machine Learning Approach. Expert Syst. Appl. 2022,197, 116817. [CrossRef] 99. Li, G.; Li, B.; Li, C.; Wang, S. State-of-Health Rapid Estimation for Lithium-Ion Battery Based on an Interpretable Stacking Ensemble Model with Short-Term Voltage Profiles. Energy 2023,263, 126064. [CrossRef] 100. Lundberg, S.M.; Lee, S.-I. A Unified Approach to Interpreting Model Predictions. In Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, CA, USA, 4–9 December 2017; Guyon, I., Von Luxburg, U., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R., Eds.; Curran Associates, Inc.: Brooklyn, NY, USA, 2017; Volume 30. 101. Lundberg, S.M.; Erion, G.; Chen, H.; DeGrave, A.; Prutkin, J.M.; Nair, B.; Katz, R.; Himmelfarb, J.; Bansal, N.; Lee, S.-I. From Local Explanations to Global Understanding with Explainable AI for Trees. Nat. Mach. Intell. 2020,2, 56–67. [CrossRef] 102. Attia, P.M.; Severson, K.A.; Witmer, J.D. Statistical Learning for Accurate and Interpretable Battery Lifetime Prediction. J. Electrochem. Soc. 2021,168, 090547. [CrossRef] 103. Lee, G.; Kwon, D.; Lee, C. A Convolutional Neural Network Model for SOH Estimation of Li-Ion Batteries with Physical Interpretability. Mech. Syst. Signal Process. 2023,188, 110004. [CrossRef] 104. Zhang, H.; Su, Y.; Altaf, F.; Wik, T.; Gros, S. Interpretable Battery Cycle Life Range Prediction Using Early Cell Degradation Data. IEEE Trans. Transp. Electrif. 2023,9, 2669–2682. [CrossRef] Energies 2025,18, 746 65 of 77 105. Pang, X.; Yang, W.; Wang, C.; Fan, H.; Wang, L.; Li, J.; Zhong, S.; Zheng, W.; Zou, H.; Chen, S.; et al. A Novel Hybrid Model for Lithium-Ion Batteries Lifespan Prediction with High Accuracy and Interpretability. J. Energy Storage 2023,61, 106728. [CrossRef] 106. Kong, D.; Wang, S.; Ping, P. State-of-Health Estimation and Remaining Useful Life for Lithium-Ion Battery Based on Deep Learning with Bayesian Hyperparameter Optimization. Int. J. Energy Res. 2022,46, 6081–6098. [CrossRef] 107. Reddy, M.C.; Viswanath, M.N.; Varma, P.S. Remaining Useful Life Prediction of Lithium Ion Battery Using Tree Based Pipe-Line Optimization Tool. J. Adv. Res. Dyn. Control Syst. 2020,12, 1955–1960. [CrossRef] 108. Shu, X.; Li, G.; Shen, J.; Lei, Z.; Chen, Z.; Liu, Y. A Uniform Estimation Framework for State of Health of Lithium-Ion Batteries Considering Feature Extraction and Parameters Optimization. Energy 2020,204, 117957. [CrossRef] 109. Couture, J.; Lin, X. Novel Image-Based Rapid RUL Prediction for Li-Ion Batteries Using a Capsule Network and Transfer Learning. IEEE Trans. Transp. Electrif. 2023,9, 958–967. [CrossRef] 110. Couture, J.; Lin, X. Imageand Health Indicator-Based Transfer Learning Hybridization for Battery RUL Prediction. Eng. Appl. Artif. Intell. 2022,114, 105120. [CrossRef] 111. He, N.; Wang, Q.; Lu, Z.; Chai, Y.; Yang, F. Early Prediction of Battery Lifetime Based on Graphical Features and Convolutional Neural Networks. Appl. Energy 2024,353, 122048. [CrossRef] 112. Zhao, X.; He, H.; Li, J.; Wei, Z.; Huang, R.; Shi, M. From Grayscale Image to Battery Aging Awareness—A New Battery Capacity Estimation Model with Computer Vision Approach. IEEE Trans. Ind. Inform. 2023,19, 8965–8975. [CrossRef] 113. Li, Y.; Li, K.; Liu, X.; Zhang, L. Fast Battery Capacity Estimation Using Convolutional Neural Networks. Trans. Inst. Meas. Control 2020, 0142331220966425. [CrossRef] 114. Teng, J.-H.; Chen, R.-J.; Lee, P.-T.; Hsu, C.-W. Accurate and Efficient SOH Estimation for Retired Batteries. Energies 2023,16, 1240. [CrossRef] 115. Huang, C.; Li, N. Fast Health State Estimation of Lead–Acid Batteries Based on Multi-Time Constant Current Charging Curve. Electronics 2023,12, 4552. [CrossRef] 116. Lu, S.; Yin, Z.; Liao, S.; Yang, B.; Liu, S.; Liu, M.; Yin, L.; Zheng, W. An Asymmetric Encoder–Decoder Model for Zn-Ion Battery Lifetime Prediction. Energy Rep. 2022,8, 33–50. [CrossRef] 117. Dubarry, M.; Beck, D. Big Data Training Data for Artificial Intelligence-Based Li-Ion Diagnosis and Prognosis. J. Power Sources 2020,479, 228806. [CrossRef] 118. Rauf, H.; Khalid, M.; Arshad, N. Machine Learning in State of Health and Remaining Useful Life Estimation: Theoretical and Technological Development in Battery Degradation Modelling. Renew. Sustain. Energy Rev. 2022,156, 111903. [CrossRef] 119. Shu, X.; Shen, S.; Shen, J.; Zhang, Y.; Li, G.; Chen, Z.; Liu, Y. State of Health Prediction of Lithium-Ion Batteries Based on Machine Learning: Advances and Perspectives. iScience 2021,24, 103265. [CrossRef] 120. Wang, S.; Jin, S.; Deng, D.; Fernandez, C. A Critical Review of Online Battery Remaining Useful Lifetime Prediction Methods. Front. Mech. Eng. 2021,7, 719718. [CrossRef] 121. Toughzaoui, Y.; Toosi, S.B.; Chaoui, H.; Louahlia, H.; Petrone, R.; Le Masson, S.; Gualous, H. State of Health Estimation and Remaining Useful Life Assessment of Lithium-Ion Batteries: A Comparative Study. J. Energy Storage 2022,51, 104520. [CrossRef] 122. Sharma, P.; Bora, B.J. A Review of Modern Machine Learning Techniques in the Prediction of Remaining Useful Life of Lithium-Ion Batteries. Batteries 2023,9, 13. [CrossRef] 123. Jin, S.; Sui, X.; Huang, X.; Wang, S.; Teodorescu, R.; Stroe, D.-I. Overview of Machine Learning Methods for Lithium-Ion Battery Remaining Useful Lifetime Prediction. Electronics 2021,10, 3126. [CrossRef] 124. Ren, Z.; Du, C. A Review of Machine Learning State-of-Charge and State-of-Health Estimation Algorithms for Lithium-Ion Batteries. Energy Rep. 2023,9, 2993–3021. [CrossRef] 125. Guo, W.; Sun, Z.; Vilsen, S.B.; Meng, J.; Stroe, D.I. Review of “Grey Box” Lifetime Modeling for Lithium-Ion Battery: Combining Physics and Data-Driven Methods. J. Energy Storage 2022,56, 105992. [CrossRef] 126. Wang, F.; Zhao, Z.; Zhai, Z.; Shang, Z.; Yan, R.; Chen, X. Explainability-Driven Model Improvement for SOH Estimation of Lithium-Ion Battery. Reliab. Eng. Syst. Saf. 2023,232, 109046. [CrossRef] 127. Zhou, L.; Lai, X.; Li, B.; Yao, Y.; Yuan, M.; Weng, J.; Zheng, Y. State Estimation Models of Lithium-Ion Batteries for Battery Management System: Status, Challenges, and Future Trends. Batteries 2023,9, 131. [CrossRef] 128. Shah, A.; Shah, K.; Shah, C.; Shah, M. State of Charge, Remaining Useful Life and Knee Point Estimation Based on Artificial Intelligence and Machine Learning in Lithium-Ion EV Batteries: A Comprehensive Review. Renew. Energy Focus 2022,42, 146–164. [CrossRef] 129. Li, X.; Yu, D.; Søren Byg, V.; Daniel Ioan, S. The Development of Machine Learning-Based Remaining Useful Life Prediction for Lithium-Ion Batteries. J. Energy Chem. 2023,82, 103–121. [CrossRef] 130. Raoofi, T.; Yildiz, M. Comprehensive Review of Battery State Estimation Strategies Using Machine Learning for Battery Management Systems of Aircraft Propulsion Batteries. J. Energy Storage 2023,59, 106486. [CrossRef] 131. Xiao, Y.; Wen, J.; Yao, L.; Zheng, J.; Fang, Z.; Shen, Y. A Comprehensive Review of the Lithium-Ion Battery State of Health Prognosis Methods Combining Aging Mechanism Analysis. J. Energy Storage 2023,65, 107347. [CrossRef] Energies 2025,18, 746 66 of 77 132. He, W.; Li, Z.; Liu, T.; Liu, Z.; Guo, X.; Du, J.; Li, X.; Sun, P.; Ming, W. Research Progress and Application of Deep Learning in Remaining Useful Life, State of Health and Battery Thermal Management of Lithium Batteries. J. Energy Storage 2023,70, 107868. [CrossRef] 133. Beltran, H.; Sansano, E.; Pecht, M. Machine Learning Techniques Suitability to Estimate the Retained Capacity in Lithium-Ion Batteries from Partial Charge/Discharge Curves. J. Energy Storage 2023,59, 106346. [CrossRef] 134. Zraibi, B.; Mansouri, M.; Loukili, S.E. Comparing Deep Learning Methods to Predict the Remaining Useful Life of Lithium-Ion Batteries. Mater. Today Proc. 2022,62, 6298–6304. [CrossRef] 135. Huang, Z.; Sugiarto, L.; Lu, Y.-C. Feature–Target Pairing in Machine Learning for Battery Health Diagnosis and Prognosis: A Critical Review. EcoMat 2023,5, e12345. [CrossRef] 136. Zhou, Y.; Chang, Y.; Wang, Y.; Li, R. Research on Methods for Extracting Aging Characteristics and Health Status of Lithium-Ion Batteries Based on Small Samples. J. Renew. Sustain. Energy 2022,14, 24101. [CrossRef] 137. Das, K.; Kumar, R. Electric Vehicle Battery Capacity Degradation and Health Estimation Using Machine-Learning Techniques: A Review. Clean Energy 2023,7, 1268–1281. [CrossRef] 138. Wang, F.; Zhai, Z.; Liu, B.; Zheng, S.; Zhao, Z.; Chen, X. Open Access Dataset, Code Library and Benchmarking Deep Learning Approaches for State-of-Health Estimation of Lithium-Ion Batteries. J. Energy Storage 2024,77, 109884. [CrossRef] 139. Orzech, G. Prognostics Center of Excellence Data Set Repository; NASA: Washington, DC, USA, 2022. 140. Duan, W.; Song, S.; Xiao, F.; Chen, Y.; Peng, S.; Song, C. Battery SOH Estimation and RUL Prediction Framework Based on Variable Forgetting Factor Online Sequential Extreme Learning Machine and Particle Filter. J. Energy Storage 2023,65, 107322. [CrossRef] 141. Yao, L.; Wen, J.; Xu, S.; Zheng, J.; Hou, J.; Fang, Z.; Xiao, Y. State of Health Estimation Based on the Long Short-Term Memory Network Using Incremental Capacity and Transfer Learning. Sensors 2022,22, 7835. [CrossRef] [PubMed] 142. Yang, X.; Ma, B.; Xie, H.; Wang, W.; Zou, B.; Liang, F.; Hua, X.; Liu, X.; Chen, S. Lithium-Ion Battery State of Health Estimation with Multi-Feature Collaborative Analysis and Deep Learning Method. Batteries 2023,9, 120. [CrossRef] 143. Peng, S.; Sun, Y.; Liu, D.; Yu, Q.; Kan, J.; Pecht, M. State of Health Estimation of Lithium-Ion Batteries Based on Multi-Health Features Extraction and Improved Long Short-Term Memory Neural Network. Energy 2023,282, 128956. [CrossRef] 144. Jorge, I.; Mesbahi, T.; Samet, A.; Boné, R. Time Series Feature Extraction for Lithium-Ion Batteries State-Of-Health Prediction. J. Energy Storage 2023,59, 106436. [CrossRef] 145. Zhang, C.; Wang, S.; Yu, C.; Xie, Y.; Fernandez, C. Improved Particle Swarm Optimization-Extreme Learning Machine Modeling Strategies for the Accurate Lithium-Ion Battery State of Health Estimation and High-Adaptability Remaining Useful Life Prediction. J. Electrochem. Soc. 2022,169, 080520. [CrossRef] 146. Maleki, S.; Ray, B.; Hagh, M.T. Hybrid Framework for Predicting and Forecasting State of Health of Lithium-Ion Batteries in Electric Vehicles. Sustain. Energy Grids Netw. 2022,30, 100603. [CrossRef] 147. Wu, W.; Lu, S. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Data Preprocessing and Improved ELM. IEEE Trans. Instrum. Meas. 2023,72, 1–14. [CrossRef] 148. Zhang, Y.; Ma, H.; Wang, S.; Li, S.; Guo, R. Indirect Prediction of Remaining Useful Life for Lithium-Ion Batteries Based on Improved Multiple Kernel Extreme Learning Machine. J. Energy Storage 2023,64, 107181. [CrossRef] 149. Sun, C.; Qu, A.; Zhang, J.; Shi, Q.; Jia, Z. Remaining Useful Life Prediction for Lithium-Ion Batteries Based on Improved Variational Mode Decomposition and Machine Learning Algorithm. Energies 2022,16, 313. [CrossRef] 150. Tang, X.; Wan, H.; Wang, W.; Gu, M.; Wang, L.; Gan, L. Lithium-Ion Battery Remaining Useful Life Prediction Based on Hybrid Model. Sustainability 2023,15, 6261. [CrossRef] 151. Xue, K.; Yang, J.; Yang, M.; Wang, D. An Improved Generic Hybrid Prognostic Method for RUL Prediction Based on PF-LSTM Learning. IEEE Trans. Instrum. Meas. 2023,72, 1–21. [CrossRef] 152. Tian, A.; Yang, C.; Gao, Y.; Li, T.; Wang, L.; Chang, C.; Jiang, J. A State of Health Estimation Method of Lithium-Ion Batteries Based on DT-IC-V Health Features Extracted from Partial Charging Segment. Int. J. Green Energy 2023,20, 997–1011. [CrossRef] 153. Guo, Y.; Yang, D.; Zhao, K.; Wang, K. State of Health Estimation for Lithium-Ion Battery Based on Bi-Directional Long Short-Term Memory Neural Network and Attention Mechanism. Energy Rep. 2022,8, 208–215. [CrossRef] 154. Pugalenthi, K.; Park, H.; Hussain, S.; Raghavan, N. Remaining Useful Life Prediction of Lithium-Ion Batteries Using Neural Networks with Adaptive Bayesian Learning. Sensors 2022,22, 3803. [CrossRef] [PubMed] 155. Hell, S.M.; Kim, C.D. Development of a Data-Driven Method for Online Battery Remaining-Useful-Life Prediction. Batteries 2022, 8, 192. [CrossRef] 156. Zhu, C.; Gao, M.; He, Z.; Wu, H.; Sun, C.; Zhang, Z.; Bao, Z. State of Health Prediction for Li-Ion Batteries with End-to-End Deep Learning. J. Energy Storage 2023,65, 107218. [CrossRef] 157. Bao, X.; Chen, L.; Lopes, A.M.; Li, X.; Xie, S.; Li, P.; Chen, Y. Hybrid Deep Neural Network with Dimension Attention for State-of-Health Estimation of Lithium-Ion Batteries. Energy 2023,278, 127734. [CrossRef] Energies 2025,18, 746 67 of 77 158. Guo, X.; Wang, K.; Yao, S.; Fu, G.; Ning, Y. RUL Prediction of Lithium Ion Battery Based on CEEMDAN-CNN BiLSTM Model. Energy Rep. 2023,9, 1299–1306. [CrossRef] 159. Sheng, H.; Zhou, Y.; Bai, L.; Shi, L. Transfer State of Health Estimation Based on Cross-Manifold Embedding. J. Energy Storage 2022,47, 103555. [CrossRef] 160. Liu, R. Remaining Useful Life Prediction of Lithium-Ion Batteries Using Multiple Kernel Extreme Learning Machine. Recent Adv. Comput. Sci. Commun. 2022,15, 715–721. [CrossRef] 161. Wu, J.; Cheng, X.; Huang, H.; Fang, C.; Zhang, L.; Zhao, X.; Zhang, L.; Xing, J. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on PSO-RF Algorithm. Front. Energy Res. 2023,10, 937035. [CrossRef] 162. Sahoo, S.; Hariharan, K.S.; Agarwal, S.; Swernath, S.B.; Bharti, R.; Han, S.; Lee, S. Transfer Learning Based Generalized Framework for State of Health Estimation of Li-Ion Cells. Sci. Rep. 2022,12, 13173. [CrossRef] [PubMed] 163. Li, N.; He, F.; Ma, W.; Wang, R.; Jiang, L.; Zhang, X. An Indirect State-of-Health Estimation Method Based on Improved Genetic and Back Propagation for Online Lithium-Ion Battery Used in Electric Vehicles. IEEE Trans. Veh. Technol. 2022,71, 12682–12690. [CrossRef] 164. Jiang, J.; Zhao, S.; Zhang, C. State-of-Health Estimate for the Lithium-Ion Battery Using Chi-Square and ELM-LSTM. World Electr. Veh. J. 2021,12, 228. [CrossRef] 165. El-Dalahmeh, M.; Al-Greer, M.; El-Dalahmeh, M.; Bashir, I. Capacity Estimation of Lithium-Ion Batteries Based on Adaptive Empirical Wavelet Transform and Long Short-Term Memory Neural Network. J. Energy Storage 2023,70, 108046. [CrossRef] 166. Tian, Y.; Wen, J.; Yang, Y.; Shi, Y.; Zeng, J. State-of-Health Prediction of Lithium-Ion Batteries Based on CNN-BiLSTM-AM. Batteries 2022,8, 155. [CrossRef] 167. Mei, P.; Karimi, H.R.; Chen, F.; Yang, S.; Huang, C.; Qiu, S. A Learning-Based Vehicle-Cloud Collaboration Approach for Joint Estimation of State-of-Energy and State-of-Health. Sensors 2022,22, 9474. [CrossRef] 168. Song, R.; Yang, L.; Chen, L.; Dong, Z. Capacity Estimation Method of Lithium-Ion Batteries Based on Deep Convolution Neural Network. Int. J. Bio-Inspired Comput. 2022,20, 119–125. [CrossRef] 169. Wang, J.; Feng, X.; Zhang, X.; Xiang, Y. Improved Modeling of Lithium-Ion Battery Capacity Degradation Using an IndividualState Training Method and Recurrent Softplus Neural Network. IEEE Access 2021,9, 7845–7855. [CrossRef] 170. Wen, L.; Bo, N.; Ye, X.; Li, X. A Novel Auto-LSTM-Based State of Health Estimation Method for Lithium-Ion Batteries. J. Electrochem. Energy Convers. Storage 2021,18, 030902. [CrossRef] 171. Wei, Y. Prediction of State of Health of Lithium-Ion Battery Using Health Index Informed Attention Model. Sensors 2023,23, 2587. [CrossRef] [PubMed] 172. Chen, C.; Wei, J.; Li, Z. Remaining Useful Life Prediction for Lithium-Ion Batteries Based on a Hybrid Deep Learning Model. Processes 2023,11, 2333. [CrossRef] 173. Zhu, T.; Wang, W.; Yu, M. A Novel Hybrid Scheme for Remaining Useful Life Prognostic Based on Secondary Decomposition, BiGRU and Error Correction. Energy 2023,276, 127565. [CrossRef] 174. Pham, T.; Le, T.; Dang, D.; Bui, H.; Pham, H.; Truong, L.; Nguyen, M.; Vo, H.; Tho, Q.T. ARNS: A Data-Driven Approach for SoH Estimation of Lithium-Ion Battery Using Nested Sequence Models with Considering Relaxation Effect. IEEE Access 2022,10, 117067–117083. [CrossRef] 175. Ang, E.Y.M.; Paw, Y.C. Linear Model for Online State of Health Estimation of Lithium-Ion Batteries Using Segmented Discharge Profiles. IEEE Trans. Transp. Electrif. 2023,9, 2464–2471. [CrossRef] 176. Hong, S.; Kang, M.; Kim, J.; Baek, J. Sequential Application of Denoising Autoencoder and Long-Short Recurrent Convolutional Network for Noise-Robust Remaining-Useful-Life Prediction Framework of Lithium-Ion Batteries. Comput. Ind. Eng. 2023, 179, 109231. [CrossRef] 177. Chen, D.; Zheng, X.; Chen, C.; Zhao, W.; Chen, D.; Zheng, X.; Chen, C.; Zhao, W. Remaining Useful Life Prediction of the Lithium-Ion Battery Based on CNN-LSTM Fusion Model and Grey Relational Analysis. Electron. Res. Arch. 2023,31, 633–655. [CrossRef] 178. Zhu, C.; Zheng, B.; He, Z.; Gao, M.; Sun, C.; Bao, Z. State of Health Estimation of Lithium-Ion Battery Using Time Convolution Memory Neural Network. Mob. Inf. Syst. 2021,2021, 4826409. [CrossRef] 179. Wei, Y.; Wu, D. State of Health and Remaining Useful Life Prediction of Lithium-Ion Batteries with Conditional Graph Convolutional Network. Expert Syst. Appl. 2024,238, 122041. [CrossRef] 180. Huang, K.; Yao, K.; Guo, Y.; Lv, Z. State of Health Estimation of Lithium-Ion Batteries Based on Fine-Tuning or Rebuilding Transfer Learning Strategies Combined with New Features Mining. Energy 2023,282, 128739. [CrossRef] 181. Zou, B.; Xiong, M.; Wang, H.; Ding, W.; Jiang, P.; Hua, W.; Zhang, Y.; Zhang, L.; Wang, W.; Tan, R. A Deep Learning Approach for State-of-Health Estimation of Lithium-Ion Batteries Based on a Multi-Feature and Attention Mechanism Collaboration. Batteries 2023,9, 329. [CrossRef] 182. Liu, H.; Li, Y.; Luo, L.; Zhang, C. A Lithium-Ion Battery Capacity and RUL Prediction Fusion Method Based on Decomposition Strategy and GRU. Batteries 2023,9, 323. [CrossRef] Energies 2025,18, 746 68 of 77 183. Yu, Z.; Liu, N.; Zhang, Y.; Qi, L.; Li, R. Battery SOH Prediction Based on Multi-Dimensional Health Indicators. Batteries 2023,9, 80. [CrossRef] 184. Ma, W.; Cai, P.; Sun, F.; Wang, X.; Gong, J. State of Health Estimation for Lithium-Ion Batteries Based on Extreme Learning Machine with Improved Blinex Loss. Int. J. Electrochem. Sci. 2022,17, 221170. [CrossRef] 185. Yuan, J.; Qin, Z.; Huang, H.; Gan, X.; Li, S.; Li, B. State of Health Estimation and Remaining Useful Life Prediction for a Lithium-Ion Battery with a Two-Layer Stacking Regressor. Energies 2023,16, 2313. [CrossRef] 186. Gao, D.; Liu, X.; Zhu, Z.; Yang, Q. A Hybrid CNN-BiLSTM Approach for Remaining Useful Life Prediction of EVs Lithium-Ion Battery. Meas. Control 2023,56, 371–383. [CrossRef] 187. Chen, L.; Xie, S.; Lopes, A.M.; Bao, X. A Vision Transformer-Based Deep Neural Network for State of Health Estimation of Lithium-Ion Batteries. Int. J. Electr. Power Energy Syst. 2023,152, 109233. [CrossRef] 188. Han, Y.; Li, C.; Zheng, L.; Lei, G.; Li, L. Remaining Useful Life Prediction of Lithium-Ion Batteries by Using a Denoising Transformer-Based Neural Network. Energies 2023,16, 6328. [CrossRef] 189. Wang, M.; Xiang, G.; Cui, L.; Zhang, Q.; Chen, J. Remaining Useful Life Distribution Prediction Framework for Lithium-Ion Battery Fused Prior Knowledge and Monitoring Data. Meas. Sci. Technol. 2023,34, 125108. [CrossRef] 190. Marri, I.; Petkovski, E.; Cristaldi, L.; Faifer, M. Comparing Machine Learning Strategies for SoH Estimation of Lithium-Ion Batteries Using a Feature-Based Approach. Energies 2023,16, 4423. [CrossRef] 191. Xie, Q.; Liu, R.; Huang, J.; Su, J. Residual Life Prediction of Lithium-Ion Batteries Based on Data Preprocessing and a Priori Knowledge-Assisted CNN-LSTM. Energy 2023,281, 128232. [CrossRef] 192. Li, Z.; Li, A.; Bai, F.; Zuo, H.; Zhang, Y. Remaining Useful Life Prediction of Lithium Battery Based on ACNN-Mogrifier LSTM-MMD. Meas. Sci. Technol. 2024,35, 016101. [CrossRef] 193. Tarar, M.O.; Naqvi, I.H.; Khalid, Z.; Pecht, M. Accurate Prediction of Remaining Useful Life for Lithium-Ion Battery Using Deep Neural Networks with Memory Features. Front. Energy Res. 2023,11, 1059701. [CrossRef] 194. Wu, J.; Liu, Z.; Zhang, Y.; Lei, D.; Zhang, B.; Cao, W. Data-Driven State of Health Estimation for Lithium-Ion Battery Based on Voltage Variation Curves. J. Energy Storage 2023,73, 109191. [CrossRef] 195. Wang, Z.; Liu, Y.; Wang, F.; Wang, H.; Su, M. Capacity and Remaining Useful Life Prediction for Lithium-Ion Batteries Based on Sequence Decomposition and a Deep-Learning Network. J. Energy Storage 2023,72, 108085. [CrossRef] 196. Chang, Y.-H.; Hsieh, Y.-C.; Chai, Y.-H.; Lin, H.-W. Remaining-Useful-Life Prediction for Li-Ion Batteries. Energies 2023,16, 3096. [CrossRef] 197. Wang, J.; Zhu, L.; Dai, H. An Efficient State-of-Health Estimation Method for Lithium-Ion Batteries Based on Feature-Importance Ranking Strategy and PSO-GRNN Algorithm. J. Energy Storage 2023,72, 108638. [CrossRef] 198. Nair, P.; Vakharia, V.; Borade, H.; Shah, M.; Wankhede, V. Predicting Li-Ion Battery Remaining Useful Life: An XDFM-Driven Approach with Explainable AI. Energies 2023,16, 5725. [CrossRef] 199. Zhang, F.; Shen, Z.; Xu, M.; Xie, Q.; Fu, Q.; Ma, R. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on TCN-DCN Fusion Model Combined with IRRS Filtering. J. Energy Storage 2023,72, 108586. [CrossRef] 200. Pan, R.; Liu, T.; Huang, W.; Wang, Y.; Yang, D.; Chen, J. State of Health Estimation for Lithium-Ion Batteries Based on Two-Stage Features Extraction and Gradient Boosting Decision Tree. Energy 2023,285, 129460. [CrossRef] 201. Zhang, Y.; Zhang, Y.; Wu, T. Estimation of State of Health Based on Charging Characteristics and Back-Propagation Neural Networks with Improved Atom Search Optimization Algorithm. Glob. Energy Interconnect. 2023,6, 228–237. [CrossRef] 202. Jafari, S.; Byun, Y.-C. A CNN-GRU Approach to the Accurate Prediction of Batteries’ Remaining Useful Life from Charging Profiles. Computers 2023,12, 219. [CrossRef] 203. Wang, Y.; He, Q.; Zhang, D.; Lu, S.; Yuan, C. Improving Li-Ion Battery Health: Predicting Remaining Useful Life Using IWBOA-ELM Algorithm. J. Energy Storage 2023,72, 108547. [CrossRef] 204. Wang, F.; Yang, Y.; Huang, T.; Xu, Y. Lifetime Prediction of Electronic Devices Based on the P-Stacking Machine Learning Model. Microelectron. Reliab. 2023,146, 115027. [CrossRef] 205. Ansari, S.; Ayob, A.; Lipu, M.S.H.; Hussain, A.; Abdolrasol, M.G.M.; Zainuri, M.A.A.M.; Saad, M.H.M. Optimized Data-Driven Approach for Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Sliding Window and Systematic Sampling. J. Energy Storage 2023,73, 109198. [CrossRef] 206. Guo, L.; He, H.; Ren, Y.; Li, R.; Jiang, B.; Gong, J. Prognostics of Lithium-Ion Batteries Health State Based on Adaptive Mode Decomposition and Long Short-Term Memory Neural Network. Eng. Appl. Artif. Intell. 2024,127, 107317. [CrossRef] 207. Zhang, Y.; Wang, Y.; Zhang, C.; Qiao, X.; Ge, Y.; Li, X.; Peng, T.; Nazir, M.S. State-of-Health Estimation for Lithium-Ion Battery via an Evolutionary Stacking Ensemble Learning Paradigm of Random Vector Functional Link and Active-State-Tracking Long–Short-Term Memory Neural Network. Appl. Energy 2024,356, 122417. [CrossRef] 208. Li, Z.; Bai, F.; Zuo, H.; Zhang, Y. Remaining Useful Life Prediction for Lithium-Ion Batteries Based on Iterative Transfer Learning and Mogrifier LSTM. Batteries 2023,9, 448. [CrossRef] Energies 2025,18, 746 69 of 77 209. Ren, Z.; Du, C.; Zhao, Y. A Novel Method for State of Health Estimation of Lithium-Ion Batteries Based on Deep Learning Neural Network and Transfer Learning. Batteries 2023,9, 585. [CrossRef] 210. Zheng, W.; Zhou, X.; Bai, C.; Zhou, D.; Fu, P. Adaptation of Deep Network in Transfer Learning for Estimating State of Health in Electric Vehicles during Operation. Batteries 2023,9, 547. [CrossRef] 211. Kuo, P.-H.; Tseng, Y.-R.; Luan, P.-C.; Yau, H.-T. Broad Transfer Learning Network Based Li-Ion Battery Lifetime Prediction Model. Energy Rep. 2023,10, 881–893. [CrossRef] 212. Bosello, M.; Falcomer, C.; Rossi, C.; Pau, G. To Charge or to Sell? EV Pack Useful Life Estimation via LSTMs, CNNs, and Autoencoders. Energies 2023,16, 2837. [CrossRef] 213. Berghout, T.; Benbouzid, M.; Amirat, Y.; Yao, G. Lithium-Ion Battery State of Health Prediction with a Robust Collaborative Augmented Hidden Layer Feedforward Neural Network Approach. IEEE Trans. Transp. Electrif. 2023,9, 4492–4502. [CrossRef] 214. Hong, J.; Chen, Y.; Chai, Q.; Lin, Q.; Wang, W. State-of-Health Estimation of Lithium-Ion Batteries Using a Novel Dual-Stage Attention Mechanism Based Recurrent Neural Network. J. Energy Storage 2023,72, 109297. [CrossRef] 215. Wang, S.; Ma, H.; Zhang, Y.; Li, S.; He, W. Remaining Useful Life Prediction Method of Lithium-Ion Batteries Is Based on Variational Modal Decomposition and Deep Learning Integrated Approach. Energy 2023,282, 128984. [CrossRef] 216. Zhang, L.; Zhang, J.; Gao, T.; Lyu, L.; Wang, L.; Shi, W.; Jiang, L.; Cai, G. Improved LSTM Based State of Health Estimation Using Random Segments of the Charging Curves for Lithium-Ion Batteries. J. Energy Storage 2023,74, 109370. [CrossRef] 217. Guo, F.; Huang, G.; Zhang, W.; Wen, A.; Li, T.; He, H.; Huang, H.; Zhu, S. Lithium Battery State-of-Health Estimation Based on Sample Data Generation and Temporal Convolutional Neural Network. Energies 2023,16, 8010. [CrossRef] 218. Liu, Y.; Sun, G.; Liu, X. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Peak Interval Features and Deep Learning. J. Energy Storage 2023,73, 109308. [CrossRef] 219. Jin, Z.; Fang, C.; Wu, J.; Li, J.; Zeng, W.; Zhao, X. Remaining Useful Life Prediction for Lithium-Ion Battery Using a Data-Driven Method. Int. J. Wirel. Mob. Comput. 2022,23, 239–249. [CrossRef] 220. Zhao, L.; Song, S.; Wang, P.; Wang, C.; Wang, J.; Guo, M. A MLP-Mixer and Mixture of Expert Model for Remaining Useful Life Prediction of Lithium-Ion Batteries. Front. Comput. Sci. 2023,18, 185329. [CrossRef] 221. Wang, X.; Hu, B.; Su, X.; Xu, L.; Zhu, D. State of Health Estimation for Lithium-Ion Batteries Using Random Forest and Gated Recurrent Unit. J. Energy Storage 2024,76, 109796. [CrossRef] 222. Chen, L.; Xie, S.; Lopes, A.M.; Li, H.; Bao, X.; Zhang, C.; Li, P. A New SOH Estimation Method for Lithium-Ion Batteries Based on Model-Data-Fusion. Energy 2024,286, 129597. [CrossRef] 223. Mazzi, Y.; Ben Sassi, H.; Errahimi, F. Lithium-Ion Battery State of Health Estimation Using a Hybrid Model Based on a Convolutional Neural Network and Bidirectional Gated Recurrent Unit. Eng. Appl. Artif. Intell. 2024,127, 107199. [CrossRef] 224. Zhigang, L.; Meng, Z.; Ruohai, D.; Peng, W.; Hui, G.; Hongxi, W. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on KS Agglomeration Function Integrating Multi-Expert Knowledge. Microelectron. Reliab. 2023,145, 114985. [CrossRef] 225. Li, C.; Han, X.; Zhang, Q.; Li, M.; Rao, Z.; Liao, W.; Liu, X.; Liu, X.; Li, G. State-of-Health and Remaining-Useful-Life Estimations of Lithium-Ion Battery Based on Temporal Convolutional Network-Long Short-Term Memory. J. Energy Storage 2023,74, 109498. [CrossRef] 226. Rincón-Maya, C.; Guevara-Carazas, F.; Hernández-Barajas, F.; Patino-Rodriguez, C.; Usuga-Manco, O. Remaining Useful Life Prediction of Lithium-Ion Battery Using ICC-CNN-LSTM Methodology. Energies 2023,16, 7081. [CrossRef] 227. Zhang, C.; Wang, S.; Yu, C.; Wang, Y.; Fernandez, C. A Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Deep Autoregressive Recurrent Neural Network Method for the Whole Life Remaining Useful Life Prediction of Lithium-Ion Batteries. Ionics 2023,29, 4337–4349. [CrossRef] 228. Cai, Y.; Li, W.; Zahid, T.; Zheng, C.; Zhang, Q.; Xu, K. Early Prediction of Remaining Useful Life for Lithium-Ion Batteries Based on CEEMDAN-Transformer-DNN Hybrid Model. Heliyon 2023,9, e17754. [CrossRef] 229. Feng, J.; Cai, F.; Li, H.; Huang, K.; Yin, H. A Data-Driven Prediction Model for the Remaining Useful Life Prediction of Lithium-Ion Batteries. Process Saf. Environ. Prot. 2023,180, 601–615. [CrossRef] 230. Xia, T.; Zhang, X.; Zhu, H.; Zhang, X.; Shen, J. An Accurate Denoising Lithium-Ion Battery Remaining Useful Life Prediction Model Based on CNN and LSTM with Self-Attention. Ionics 2023,29, 5315–5328. [CrossRef] 231. Ding, G.; Chen, H. A Novel Lithium-Ion Battery Capacity Prediction Framework Based on SVMD-AO-DELM. Signal Image Video Process. 2023,17, 3793–3801. [CrossRef] 232. Xue, J.; Ma, W.; Feng, X.; Guo, P.; Guo, Y.; Hu, X.; Chen, B. Stacking Integrated Learning Model via ELM and GRU with Mixture Correntropy Loss for Robust State of Health Estimation of Lithium-Ion Batteries. Energy 2023,284, 129279. [CrossRef] 233. Zheng, X.; Chen, D.; Wang, Y.; Zhuang, L.; Zheng, X.; Chen, D.; Wang, Y.; Zhuang, L. Remaining Useful Life Indirect Prediction of Lithium-Ion Batteries Using CNN-BiGRU Fusion Model and TPE Optimization. AIMS Energy 2023,11, 896–917. [CrossRef] 234. Geng, C.; Zhang, T.; Chen, B.; Zhou, Q. Battery State of Health Estimation Using GA-BP Neural Network on Data Feature Mining. IEICE Electron. Express 2023,20, 20230370. [CrossRef] Energies 2025,18, 746 70 of 77 235. Sun, S.; Zhang, H.; Ge, J.; Che, L. State-of-Health Estimation for Lithium-Ion Battery Using Model-Based Feature Optimization and Deep Extreme Learning Machine. J. Energy Storage 2023,72, 108732. [CrossRef] 236. Zraibi, B.; Okar, C.; Chaoui, H.; Mansouri, M. Remaining Useful Life Assessment for Lithium-Ion Batteries Using CNN-LSTMDNN Hybrid Method. IEEE Trans. Veh. Technol. 2021,70, 4252–4261. [CrossRef] 237. Zhao, S.; Zhang, C.; Wang, Y. Lithium-Ion Battery Capacity and Remaining Useful Life Prediction Using Board Learning System and Long Short-Term Memory Neural Network. J. Energy Storage 2022,52, 104901. [CrossRef] 238. Li, Y.; Sheng, H.; Cheng, Y.; Stroe, D.-I.; Teodorescu, R. State-of-Health Estimation of Lithium-Ion Batteries Based on SemiSupervised Transfer Component Analysis. Appl. Energy 2020,277, 115504. [CrossRef] 239. Guo, Y.; Yang, D.; Zhang, Y.; Wang, L.; Wang, K. Online Estimation of SOH for Lithium-Ion Battery Based on SSA-Elman Neural Network. Prot. Control Mod. Power Syst. 2022,7, 40. [CrossRef] 240. Chen, D.; Hong, W.; Zhou, X. Transformer Network for Remaining Useful Life Prediction of Lithium-Ion Batteries. IEEE Access 2022,10, 19621–19628. [CrossRef] 241. Nascimento, R.G.; Corbetta, M.; Kulkarni, C.S.; Viana, F.A.C. Hybrid Physics-Informed Neural Networks for Lithium-Ion Battery Modeling and Prognosis. J. Power Sources 2021,513, 230526. [CrossRef] 242. Li, Y.; Sheng, H.; Cheng, Y.; Kuang, H. Lithium-Ion Battery State of Health Monitoring Based on Ensemble Learning. In Proceedings of the 2019 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2019, Auckland, New Zealand, 20–23 May 2019. 243. Li, P.; Zhang, Z.; Grosu, R.; Deng, Z.; Hou, J.; Rong, Y.; Wu, R. An End-to-End Neural Network Framework for State-of-Health Estimation and Remaining Useful Life Prediction of Electric Vehicle Lithium Batteries. Renew. Sustain. Energy Rev. 2022, 156, 111843. [CrossRef] 244. Gou, B.; Xu, Y.; Feng, X. An Ensemble Learning-Based Data-Driven Method for Online State-of-Health Estimation of Lithium-Ion Batteries. IEEE Trans. Transp. Electrif. 2021,7, 422–436. [CrossRef] 245. Wang, S.; Fan, Y.; Jin, S.; Takyi-Aninakwa, P.; Fernandez, C. Improved Anti-Noise Adaptive Long Short-Term Memory Neural Network Modeling for the Robust Remaining Useful Life Prediction of Lithium-Ion Batteries. Reliab. Eng. Syst. Saf. 2023, 230, 108920. [CrossRef] 246. Catelani, M.; Ciani, L.; Fantacci, R.; Patrizi, G.; Picano, B. Remaining Useful Life Estimation for Prognostics of Lithium-Ion Batteries Based on Recurrent Neural Network. IEEE Trans. Instrum. Meas. 2021,70, 1–11. [CrossRef] 247. Khan, N.; Ullah, F.U.M.; Afnan; Ullah, A.; Lee, M.Y.; Baik, S.W. Batteries State of Health Estimation via Efficient Neural Networks with Multiple Channel Charging Profiles. IEEE Access 2021,9, 7797–7813. [CrossRef] 248. Sun, H.; Sun, J.; Zhao, K.; Wang, L.; Wang, K. Data-Driven ICA-Bi-LSTM-Combined Lithium Battery SOH Estimation. Math. Probl. Eng. 2022,2022, 9645892. [CrossRef] 249. Kim, S.; Choi, Y.Y.; Kim, K.J.; Choi, J.-I. Forecasting State-of-Health of Lithium-Ion Batteries Using Variational Long Short-Term Memory with Transfer Learning. J. Energy Storage 2021,41, 102893. [CrossRef] 250. Tang, T.; Yuan, H. A Hybrid Approach Based on Decomposition Algorithm and Neural Network for Remaining Useful Life Prediction of Lithium-Ion Battery. Reliab. Eng. Syst. Saf. 2022,217, 108082. [CrossRef] 251. Han, T.; Wang, Z.; Meng, H. End-to-End Capacity Estimation of Lithium-Ion Batteries with an Enhanced Long Short-Term Memory Network Considering Domain Adaptation. J. Power Sources 2022,520, 230823. [CrossRef] 252. Ding, P.; Liu, X.; Li, H.; Huang, Z.; Zhang, K.; Shao, L.; Abedinia, O. Useful Life Prediction Based on Wavelet Packet Decomposition and Two-Dimensional Convolutional Neural Network for Lithium-Ion Batteries. Renew. Sustain. Energy Rev. 2021,148, 111287. [CrossRef] 253. Zhang, Y.; Wik, T.; Bergström, J.; Pecht, M.; Zou, C. A Machine Learning-Based Framework for Online Prediction of Battery Ageing Trajectory and Lifetime Using Histogram Data. J. Power Sources 2022,526, 231110. [CrossRef] 254. Chinomona, B.; Chung, C.; Chang, L.-K.; Su, W.-C.; Tsai, M.-C. Long Short-Term Memory Approach to Estimate Battery Remaining Useful Life Using Partial Data. IEEE Access 2020,8, 165419–165431. [CrossRef] 255. Goh, H.H.; Lan, Z.; Zhang, D.; Dai, W.; Kurniawan, T.A.; Goh, K.C. Estimation of the State of Health (SOH) of Batteries Using Discrete Curvature Feature Extraction. J. Energy Storage 2022,50, 104646. [CrossRef] 256. Wei, M.; Gu, H.; Ye, M.; Wang, Q.; Xu, X.; Wu, C. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Monte Carlo Dropout and Gated Recurrent Unit. Energy Rep. 2021,7, 2862–2871. [CrossRef] 257. Wang, Z.; Zeng, S.; Guo, J.; Qin, T. Remaining Capacity Estimation of Lithium-Ion Batteries Based on the Constant Voltage Charging Profile. PLoS ONE 2018,13, e0200169. [CrossRef] 258. Guo, Y.; Yu, P.; Zhu, C.; Zhao, K.; Wang, L.; Wang, K. A State-of-Health Estimation Method Considering Capacity Recovery of Lithium Batteries. Int. J. Energy Res. 2022,46, 23730–23745. [CrossRef] 259. Gu, X.; See, K.W.; Li, P.; Shan, K.; Wang, Y.; Zhao, L.; Lim, K.C.; Zhang, N. A Novel State-of-Health Estimation for the Lithium-Ion Battery Using a Convolutional Neural Network and Transformer Model. Energy 2023,262, 125501. [CrossRef] Energies 2025,18, 746 71 of 77 260. Yang, Z.; Wang, Y.; Kong, C. Remaining Useful Life Prediction of Lithium-Ion Batteries Based on a Mixture of Ensemble Empirical Mode Decomposition and GWO-SVR Model. IEEE Trans. Instrum. Meas. 2021,70, 1–11. [CrossRef] 261. Kara, A. A Data-Driven Approach Based on Deep Neural Networks for Lithium-Ion Battery Prognostics. Neural Comput. Appl. 2021,33, 13525–13538. [CrossRef] 262. Song, S.; Fei, C.; Xia, H. Lithium-Ion Battery SOH Estimation Based on XGBoost Algorithm with Accuracy Correction. Energies 2020,13, 812. [CrossRef] 263. Liu, W.; Xu, Y.; Feng, X. A Hierarchical and Flexible Data-Driven Method for Online State-of-Health Estimation of Li-Ion Battery. IEEE Trans. Veh. Technol. 2020,69, 14739–14748. [CrossRef] 264. Wu, J.; Fang, L.; Dong, G.; Lin, M. State of Health Estimation of Lithium-Ion Battery with Improved Radial Basis Function Neural Network. Energy 2023,262, 125380. [CrossRef] 265. Cheng, Y.; Song, D.; Wang, Z.; Lu, C.; Zerhouni, N. An Ensemble Prognostic Method for Lithium-Ion Battery Capacity Estimation Based on Time-Varying Weight Allocation. Appl. Energy 2020,266, 114817. [CrossRef] 266. Mao, L.; Xu, J.; Chen, J.; Zhao, J.; Wu, Y.; Yao, F. A LSTM-STW and GS-LM Fusion Method for Lithium-Ion Battery RUL Prediction Based on EEMD. Energies 2020,13, 2380. [CrossRef] 267. Kohtz, S.; Xu, Y.; Zheng, Z.; Wang, P. Physics-Informed Machine Learning Model for Battery State of Health Prognostics Using Partial Charging Segments. Mech. Syst. Signal Process. 2022,172, 109002. [CrossRef] 268. Ardeshiri, R.R.; Liu, M.; Ma, C. Multivariate Stacked Bidirectional Long Short Term Memory for Lithium-Ion Battery Health Management. Reliab. Eng. Syst. Saf. 2022,224, 108481. [CrossRef] 269. Gong, D.; Gao, Y.; Kou, Y.; Wang, Y. State of Health Estimation for Lithium-Ion Battery Based on Energy Features. Energy 2022, 257, 124812. [CrossRef] 270. Wu, J.; Cui, X.; Meng, J.; Peng, J.; Lin, M. Data-Driven Transfer-Stacking-Based State of Health Estimation for Lithium-Ion Batteries. IEEE Trans. Ind. Electron. 2024,71, 604–614. [CrossRef] 271. Driscoll, L.; de la Torre, S.; Gomez-Ruiz, J.A. Feature-Based Lithium-Ion Battery State of Health Estimation with Artificial Neural Networks. J. Energy Storage 2022,50, 104584. [CrossRef] 272. Qin, P.; Zhao, L.; Liu, Z. State of Health Prediction for Lithium-Ion Battery Using a Gradient Boosting-Based Data-Driven Method. J. Energy Storage 2022,47, 103644. [CrossRef] 273. Cao, M.; Zhang, T.; Wang, J.; Liu, Y. A Deep Belief Network Approach to Remaining Capacity Estimation for Lithium-Ion Batteries Based on Charging Process Features. J. Energy Storage 2022,48, 103825. [CrossRef] 274. Wei, Y.; Wu, D. Prediction of State of Health and Remaining Useful Life of Lithium-Ion Battery Using Graph Convolutional Network with Dual Attention Mechanisms. Reliab. Eng. Syst. Saf. 2023,230, 108947. [CrossRef] 275. Chen, Z.; Chen, L.; Shen, W.; Xu, K. Remaining Useful Life Prediction of Lithium-Ion Battery via a Sequence Decomposition and Deep Learning Integrated Approach. IEEE Trans. Veh. Technol. 2022,71, 1466–1479. [CrossRef] 276. Ma, B.; Yang, S.; Zhang, L.; Wang, W.; Chen, S.; Yang, X.; Xie, H.; Yu, H.; Wang, H.; Liu, X. Remaining Useful Life and State of Health Prediction for Lithium Batteries Based on Differential Thermal Voltammetry and a Deep-Learning Model. J. Power Sources 2022,548, 232030. [CrossRef] 277. Ruan, H.; Wei, Z.; Shang, W.; Wang, X.; He, H. Artificial Intelligence-Based Health Diagnostic of Lithium-Ion Battery Leveraging Transient Stage of Constant Current and Constant Voltage Charging. Appl. Energy 2023,336, 120751. [CrossRef] 278. Tang, A.; Jiang, Y.; Yu, Q.; Zhang, Z. A Hybrid Neural Network Model with Attention Mechanism for State of Health Estimation of Lithium-Ion Batteries. J. Energy Storage 2023,68, 107734. [CrossRef] 279. Lin, M.; You, Y.; Wang, W.; Wu, J. Battery Health Prognosis with Gated Recurrent Unit Neural Networks and Hidden Markov Model Considering Uncertainty Quantification. Reliab. Eng. Syst. Saf. 2023,230, 108978. [CrossRef] 280. Bockrath, S.; Lorentz, V.; Pruckner, M. State of Health Estimation of Lithium-Ion Batteries with a Temporal Convolutional Neural Network Using Partial Load Profiles. Appl. Energy 2023,329, 120307. [CrossRef] 281. Xu, H.; Wu, L.; Xiong, S.; Li, W.; Garg, A.; Gao, L. An Improved CNN-LSTM Model-Based State-of-Health Estimation Approach for Lithium-Ion Batteries. Energy 2023,276, 127585. [CrossRef] 282. Liu, Y.; Sun, J.; Shang, Y.; Zhang, X.; Ren, S.; Wang, D. A Novel Remaining Useful Life Prediction Method for Lithium-Ion Battery Based on Long Short-Term Memory Network Optimized by Improved Sparrow Search Algorithm. J. Energy Storage 2023, 61, 106645. [CrossRef] 283. Bamati, S.; Chaoui, H. Lithium-Ion Batteries Long Horizon Health Prognostic Using Machine Learning. IEEE Trans. Energy Convers. 2022,37, 1176–1186. [CrossRef] 284. Cao, M.; Zhang, T.; Yu, B.; Liu, Y. A Method for Interval Prediction of Satellite Battery State of Health Based on Sample Entropy. IEEE Access 2019,7, 141549–141561. [CrossRef] 285. Zhang, Z.; Li, L.; Li, X.; Hu, Y.; Huang, K.; Xue, B.; Wang, Y.; Yu, Y. State-of-Health Estimation for the Lithium-Ion Battery Based on Gradient Boosting Decision Tree with Autonomous Selection of Excellent Features. Int. J. Energy Res. 2022,46, 1756–1765. [CrossRef] Energies 2025,18, 746 72 of 77 286. Su, C.; Chen, H.; Wen, Z. Prediction of Remaining Useful Life for Lithium-Ion Battery with Multiple Health Indicators. Eksploat. i Niezawodn. 2021,23, 176–183. [CrossRef] 287. Chen, Z.; Xue, Q.; Wu, Y.; Shen, S.; Zhang, Y.; Shen, J. Capacity Prediction and Validation of Lithium-Ion Batteries Based on Long Short-Term Memory Recurrent Neural Network. IEEE Access 2020,8, 172783–172798. [CrossRef] 288. He, J.; Tian, Y.; Wu, L. A Hybrid Data-Driven Method for Rapid Prediction of Lithium-Ion Battery Capacity. Reliab. Eng. Syst. Saf. 2022,226, 108674. [CrossRef] 289. Lin, C.; Xu, J.; Hou, J.; Liang, Y.; Mei, X. Ensemble Method with Heterogeneous Models for Battery State-of-Health Estimation. IEEE Trans. Ind. Inform. 2023,19, 10160–10169. [CrossRef] 290. Jia, J.; Wang, K.; Shi, Y.; Wen, J.; Pang, X.; Zeng, J. A Multi-Scale State of Health Prediction Framework of Lithium-Ion Batteries Considering the Temperature Variation during Battery Discharge. J. Energy Storage 2021,42, 103076. [CrossRef] 291. Vakharia, V.; Shah, M.; Nair, P.; Borade, H.; Sahlot, P.; Wankhede, V. Estimation of Lithium-Ion Battery Discharge Capacity by Integrating Optimized Explainable-AI and Stacked LSTM Model. Batteries 2023,9, 125. [CrossRef] 292. Li, D.; Yang, L. Remaining Useful Life Prediction of Lithium Battery Based on Sequential CNN–LSTM Method. J. Electrochem. Energy Convers. Storage 2021,18, 041005. [CrossRef] 293. Fan, Z.; Zi-xuan, X.; Ming-hu, W. State of Health Estimation for Li-Ion Battery Using Characteristic Voltage Intervals and Genetic Algorithm Optimized Back Propagation Neural Network. J. Energy Storage 2023,57, 106277. [CrossRef] 294. Wei, Z.; Han, X.; Li, J. State of Health Assessment for Echelon Utilization Batteries Based on Deep Neural Network Learning with Error Correction. J. Energy Storage 2022,51, 104428. [CrossRef] 295. Tang, T.; Yuan, H. An Indirect Remaining Useful Life Prognosis for Li-Ion Batteries Based on Health Indicator and Novel Artificial Neural Network. J. Energy Storage 2022,52, 104701. [CrossRef] 296. Chen, X.; Liu, Z.; Sheng, H.; Wu, K.; Mi, J.; Li, Q. Transfer Learning Based Remaining Useful Life Prediction of Lithium-Ion Battery Considering Capacity Regeneration Phenomenon. J. Energy Storage 2024,76, 109798. [CrossRef] 297. Bao, Z.; Jiang, J.; Zhu, C.; Gao, M. A New Hybrid Neural Network Method for State-of-Health Estimation of Lithium-Ion Battery. Energies 2022,15, 4399. [CrossRef] 298. Wang, Z.; Liu, N.; Chen, C.; Guo, Y. Adaptive Self-Attention LSTM for RUL Prediction of Lithium-Ion Batteries. Inf. Sci. 2023,635, 398–413. [CrossRef] 299. Yao, X.-Y.; Chen, G.; Pecht, M.; Chen, B. A Novel Graph-Based Framework for State of Health Prediction of Lithium-Ion Battery. J. Energy Storage 2023,58, 106437. [CrossRef] 300. Mao, L.; Hu, H.; Chen, J.; Zhao, J.; Qu, K.; Jiang, L. Online State-of-Health Estimation Method for Lithium-Ion Battery Based on CEEMDAN for Feature Analysis and RBF Neural Network. IEEE J. Emerg. Sel. Top. Power Electron. 2023,11, 187–200. [CrossRef] 301. Bi, J.; Lee, J.-C.; Liu, H. Performance Comparison of Long Short-Term Memory and a Temporal Convolutional Network for State of Health Estimation of a Lithium-Ion Battery Using Its Charging Characteristics. Energies 2022,15, 2448. [CrossRef] 302. Cui, Y.; Chen, Y. Prognostics of Lithium-Ion Batteries Based on Capacity Regeneration Analysis and Long Short-Term Memory Network. IEEE Trans. Instrum. Meas. 2022,71, 1–13. [CrossRef] 303. Chen, Z.; Zhang, S.; Shi, N.; Li, F.; Wang, Y.; Cui, J. Online State-of-Health Estimation of Lithium-Ion Battery Based on Relevance Vector Machine with Dynamic Integration. Appl. Soft Comput. 2022,129, 109615. [CrossRef] 304. Zhao, J.; Zhu, Y.; Zhang, B.; Liu, M.; Wang, J.; Liu, C.; Zhang, Y. Method of Predicting SOH and RUL of Lithium-Ion Battery Based on the Combination of LSTM and GPR. Sustainability 2022,14, 11865. [CrossRef] 305. Zhang, D.; Li, W.; Han, X.; Lu, B.; Zhang, Q.; Bo, C. Evolving Elman Neural Networks Based State-of-Health Estimation for Satellite Lithium-Ion Batteries. J. Energy Storage 2023,59, 106571. [CrossRef] 306. Zhu, M.; Ouyang, Q.; Wan, Y.; Wang, Z. Remaining Useful Life Prediction of Lithium-Ion Batteries: A Hybrid Approach of Grey–Markov Chain Model and Improved Gaussian Process. IEEE J. Emerg. Sel. Top. Power Electron. 2023,11, 143–153. [CrossRef] 307. Ang, E.Y.M.; Paw, Y.C. Efficient Linear Predictive Model with Short Term Features for Lithium-Ion Batteries State of Health Estimation. J. Energy Storage 2021,44, 103409. [CrossRef] 308. Ouyang, M.; Shen, P. Prediction of Remaining Useful Life of Lithium Batteries Based on WOA-VMD and LSTM. Energies 2022,15, 8918. [CrossRef] 309. Zhang, Y.; Wang, Y.; Xia, Y.; Chen, W. A Deep Learning Approach to Estimate the State of Health of Lithium-Ion Batteries under Varied and Incomplete Working Conditions. J. Energy Storage 2023,58, 106323. [CrossRef] 310. Ma, B.; Zhang, L.; Yu, H.; Zou, B.; Wang, W.; Zhang, C.; Yang, S.; Liu, X. End-Cloud Collaboration Method Enables Accurate State of Health and Remaining Useful Life Online Estimation in Lithium-Ion Batteries. J. Energy Chem. 2023,82, 1–17. [CrossRef] 311. Gao, M.; Bao, Z.; Zhu, C.; Jiang, J.; He, Z.; Dong, Z.; Song, Y. HFCM-LSTM: A Novel Hybrid Framework for State-of-Health Estimation of Lithium-Ion Battery. Energy Rep. 2023,9, 2577–2590. [CrossRef] 312. Li, L.; Li, Y.; Mao, R.; Li, L.; Hua, W.; Zhang, J. Remaining Useful Life Prediction for Lithium-Ion Batteries with a Hybrid Model Based on TCN-GRU-DNN and Dual Attention Mechanism. IEEE Trans. Transp. Electrif. 2023,9, 4726–4740. [CrossRef] Energies 2025,18, 746 73 of 77 313. Xu, J.; Liu, B.; Zhang, G.; Zhu, J. State-of-Health Estimation for Lithium-Ion Batteries Based on Partial Charging Segment and Stacking Model Fusion. Energy Sci. Eng. 2023,11, 383–397. [CrossRef] 314. Zou, L.; Wen, B.; Wei, Y.; Zhang, Y.; Yang, J.; Zhang, H. Online Prediction of Remaining Useful Life for Li-Ion Batteries Based on Discharge Voltage Data. Energies 2022,15, 2237. [CrossRef] 315. Yang, Y. A Machine-Learning Prediction Method of Lithium-Ion Battery Life Based on Charge Process for Different Applications. Appl. Energy 2021,292, 116897. [CrossRef] 316. Deng, Z.; Lin, X.; Cai, J.; Hu, X. Battery Health Estimation with Degradation Pattern Recognition and Transfer Learning. J. Power Sources 2022,525, 231027. [CrossRef] 317. Zhang, Y.; Peng, Z.; Guan, Y.; Wu, L. Prognostics of Battery Cycle Life in the Early-Cycle Stage Based on Hybrid Model. Energy 2021,221, 119901. [CrossRef] 318. Hsu, C.-W.; Xiong, R.; Chen, N.-Y.; Li, J.; Tsou, N.-T. Deep Neural Network Battery Life and Voltage Prediction by Using Data of One Cycle Only. Appl. Energy 2022,306, 118134. [CrossRef] 319. Zhang, Q.; Yang, L.; Guo, W.; Qiang, J.; Peng, C.; Li, Q.; Deng, Z. A Deep Learning Method for Lithium-Ion Battery Remaining Useful Life Prediction Based on Sparse Segment Data via Cloud Computing System. Energy 2022,241, 122716. [CrossRef] 320. Herring, P.; Balaji Gopal, C.; Aykol, M.; Montoya, J.H.; Anapolsky, A.; Attia, P.M.; Gent, W.; Hummelshøj, J.S.; Hung, L.; Kwon, H.-K.; et al. BEEP: A Python Library for Battery Evaluation and Early Prediction. SoftwareX 2020,11, 100506. [CrossRef] 321. Sanz-Gorrachategui, I.; Pastor-Flores, P.; Pajovic, M.; Wang, Y.; Orlik, P.V.; Bernal-Ruiz, C.; Bono-Nuez, A.; Artal-Sevil, J.S. Remaining Useful Life Estimation for LFP Cells in Second-Life Applications. IEEE Trans. Instrum. Meas. 2021,70, 1–10. [CrossRef] 322. Ma, G.; Xu, S.; Yang, T.; Du, Z.; Zhu, L.; Ding, H.; Yuan, Y. A Transfer Learning-Based Method for Personalized State of Health Estimation of Lithium-Ion Batteries. IEEE Trans. Neural Netw. Learn. Syst. 2024,35, 759–769. [CrossRef] 323. Zhang, Y.; Zhao, M. Cloud-Based in-Situ Battery Life Prediction and Classification Using Machine Learning. Energy Storage Mater. 2023,57, 346–359. [CrossRef] 324. Guo, F.; Wu, X.; Liu, L.; Ye, J.; Wang, T.; Fu, L.; Wu, Y. Prediction of Remaining Useful Life and State of Health of Lithium Batteries Based on Time Series Feature and Savitzky-Golay Filter Combined with Gated Recurrent Unit Neural Network. Energy 2023, 270, 126880. [CrossRef] 325. Che, Y.; Stroe, D.-I.; Hu, X.; Teodorescu, R. Semi-Supervised Self-Learning-Based Lifetime Prediction for Batteries. IEEE Trans. Ind. Inform. 2023,19, 6471–6481. [CrossRef] 326. Xu, L.; Deng, Z.; Xie, Y.; Lin, X.; Hu, X. A Novel Hybrid Physics-Based and Data-Driven Approach for Degradation Trajectory Prediction in Li-Ion Batteries. IEEE Trans. Transp. Electrif. 2023,9, 2628–2644. [CrossRef] 327. Chen, D.; Zhang, W.; Zhang, C.; Sun, B.; Cong, X.; Wei, S.; Jiang, J. A Novel Deep Learning-Based Life Prediction Method for Lithium-Ion Batteries with Strong Generalization Capability under Multiple Cycle Profiles. Appl. Energy 2022,327, 120114. [CrossRef] 328. Fei, Z.; Zhang, Z.; Yang, F.; Tsui, K.-L. A Deep Attention-Assisted and Memory-Augmented Temporal Convolutional Network Based Model for Rapid Lithium-Ion Battery Remaining Useful Life Predictions with Limited Data. J. Energy Storage 2023, 62, 106903. [CrossRef] 329. Gong, D.; Gao, Y.; Kou, Y.; Wang, Y. Early Prediction of Cycle Life for Lithium-Ion Batteries Based on Evolutionary Computation and Machine Learning. J. Energy Storage 2022,51, 104376. [CrossRef] 330. Zhang, S.; Liu, Z.; Su, H. State of Health Estimation for Lithium-Ion Batteries on Few-Shot Learning. Energy 2023,268, 126726. [CrossRef] 331. Zhou, K.Q.; Qin, Y.; Yuen, C. Transfer-Learning-Based State-of-Health Estimation for Lithium-Ion Battery with Cycle Synchronization. IEEE/ASME Trans. Mechatron. 2023,28, 692–702. [CrossRef] 332. Kim, S.; Jung, H.; Lee, M.; Choi, Y.Y.; Choi, J.-I. Model-Free Reconstruction of Capacity Degradation Trajectory of Lithium-Ion Batteries Using Early Cycle Data. eTransportation 2023,17, 100243. [CrossRef] 333. Xu, Q.; Wu, M.; Khoo, E.; Chen, Z.; Li, X. A Hybrid Ensemble Deep Learning Approach for Early Prediction of Battery Remaining Useful Life. IEEE/CAA J. Autom. Sin. 2023,10, 177–187. [CrossRef] 334. Celik, B.; Sandt, R.; dos Santos, L.C.P.; Spatschek, R. Prediction of Battery Cycle Life Using Early-Cycle Data, Machine Learning and Data Management. Batteries 2022,8, 266. [CrossRef] 335. Chen, Z.; Chen, L.; Ma, Z.; Xu, K.; Zhou, Y.; Shen, W. Joint Modeling for Early Predictions of Li-Ion Battery Cycle Life and Degradation Trajectory. Energy 2023,277, 127633. [CrossRef] 336. Ansari, S.; Ayob, A.; Hossain Lipu, M.S.; Hussain, A.; Md Saad, M.H. Jellyfish Optimized Recurrent Neural Network for State of Health Estimation of Lithium-Ion Batteries. Expert Syst. Appl. 2024,238, 121904. [CrossRef] 337. Amogne, Z.E.; Wang, F.-K.; Chou, J.-H. Transfer Learning Based on Transferability Measures for State of Health Prediction of Lithium-Ion Batteries. Batteries 2023,9, 280. [CrossRef] 338. Kheirkhah-Rad, E.; Parvareh, A.; Moeini-Aghtaie, M.; Dehghanian, P. A Data-Driven State-of-Health Estimation Model for Lithium-Ion Batteries Using Referenced-Based Charging Time. IEEE Trans. Power Deliv. 2023,38, 3406–3416. [CrossRef]